Oliver Norton

Andrew Woods, Oliver Norton

ok thanks

I think it's November 9th. I wouldn't be the person to ask though.

olvier when is our ucas dedline?

streaming league

http://www.twitch.tv/awoods321 my stream

lol

Yep, apparently some of the original papers got lost in Amsterdam and these were replacement papers.

was it scaled up a shit tonne?

96 but I got something like 50/75

what'd you get?

and I ended up having to do it

yes

in regard to difficulty?

k I was asking because the june 2013 one is probably the worst C3 paper ever

i think she will do a pick and mix

nope

Do you know which year the mock is?

I downloaded it this weekend, it's really addictive

im still super into league of legends

why

no

This is a bit off topic but have you ever played dwarf fortress?

cheers

kk

No identities I think but in terms of differentiation, everything but sin(x) and cos(x)

how much of the trig identities do you have on the formula sheet

I think you have it on the formula sheet anyway

which is ok

weve used the chain rule

so i honestly couldnt tell you

we havent went over it in calss

Do you use (V(du/dx)-U(dv/dx))/V^2

the quotient rule confuses me

i know them but they werent really explained to us

Do you know the product rule and quotient rule?

You'll begin to love differentiation once you've done C4

and we have a mock tomorrow and we havent even went over differentiation

but i hate differentiation atm

yeah, that sucks

Nothing was really hard in particular but finding identities was a bitch

what part of c3 did you find most difficult

question

hey

hey oliver

nm

hi oliver, whats the domain for our school email

i dont liike graham he confuses me

It doesn't really work as infinity isn't actually a number, it's more of a concept. Btw, if you want a huge actual number, google Graham's number.

http://images.4chan.org/b/src/1375402804707.jpg

i browse /mu/ but that's about it

i dont really like either of the sites that much anymore

oh dear

The raid only started 20 minutes ago

lol

I think operation 4chan has a delete system32 .bat

And yes they are thinking of doing a ddos

Apparently a few days ago a 4chan user posted an IP on 9gag to ddos saying it was 4chan when it was really 9gag

by 4chan

is this to get 9gag sut down thwen

check it out (definitely NSFW) http://9gag.com/fresh

no

Hey, have you heard of operation 9gag?

this, we must use this http://www.youtube.com/watch?v=U6xJfP7-HCc

yeah ive just finished vomiting

http://www.youtube.com/watch?v=_Gz9um3wV1o

Hey have you seen this yet?

ah right

They do kind of experimental hip-hop. It's pretty interesting.

nope

Have you ever heard of Death Grips?

yeah

We had switched to Sky because of unlimited downloads but now I can't really download anything without torrenting and letting it sit for a day

feels bad

lol

max

I'm not kidding 40kB download

40 meg download bro

bt infinity master race

lololol

I have the interconnection of a back water hick village so unfortunately I can't even play starcraft

up intos silver bro

just listen to the whole album while playing ranked 3s

tbh im a fan of it all

I didn't really like Game of Love though

but it works alright

the singers a bit poofy

Thoughts on Touch?

shit be cash

fkin love it

all of it

have you heard of any of Daft Punk's new album?

i dont even http://images.4chan.org/b/src/1369785176068.jpg

Well obviously the Jews did it

Last couple of panels of part 2

wierded out

got a bit confused and

sort of

lerl http://images.4chan.org/b/src/1369784305203.gif

did you read the Sims 3 one?

same i think

I'm calling fake though on this http://imgur.com/a/od0YH#0

http://imgur.com/3ClsuZB

not the mum one but this is hilarious

one sec

kk

links or?

no?

did you hear about the thread where OP fucks his mum?

and op conceeded

he took over the thread by accident

lol

people kept asking him if he did motion capture for the simpsons lol

no

also, did you see that 3 fingered guy on /b/ ysterday?

and part 2 http://imgur.com/a/7s7Nc

gimme a sec then

lol ever seen this http://imgur.com/a/EChIV its a long read

sooo good lol

seen it

enjoy this http://images.4chan.org/b/src/1369778063000.jpg

hi oliver

http://www.youtube.com/watch?feature=player_embedded&v=Er2YBunDz0s

R u ther?

Stephen Norton, Oliver Norton

I'll give it a go (when the Nexus charges!)

Now the panning and zooming should work on your tablet

http://asoiaftree.org/

asoiaftree.org Now I have panning and zooming.

Now I've got a search bar working http://asoiaftree.org/full_tree/search?query=stark

Ahhh!

Also. The reason why that particular song sounded like an 80's Meatloaf song is because they worked with one of his producers.

Adding a dropdown menu is pretty much impossible. So, I've made a compromise and made clicking the box lead to the character's family tree and clicking the text lead to the wiki page.

Haha!

Meh

This is becoming a real labour of love for you.

OK.

I'll change it tomorrow so that they refer to names rather than id numbers. The reason why some are broken as because they're very minor characters.

All of them from 1 to 300 or so. There are a few that are broken.

Found Freys and a Lanister

Got it - all Starks so far?

A few of them will generate errors at the moment

No navigation yet but replace id=1 with any number between 1 and 300

Update on the website. Now I've got characters' ancestors and descendants displayed http://asoiaftree.org/character/display?id=1

We'll talk about it tomorrow - your search function will need to be an intelligent one.

Thanks. What I've got left is to add a search to look up individual characters and to pretty it up.

Very good - all the links work well.

http://asoiaftree.org/ Check it out

Yes, they were pretty good.

Hi Oller - did your Mum tell you about Catherine's GCSEs?

Yes - great idea though.

Unfortunately I was beaten to it http://www.gizmag.com/fretlight-led-guitar-tuition-system/23048/

Thanks Oliver - you showed me the censorship article before.

https://en.wikipedia.org/wiki/Field-programmable_gate_array

https://github.com/NARKOZ/hacker-scripts

Thanks Oller

Here's those results on censorship of uk universities I was telling you about. http://www.spiked-online.com/free-speech-university-rankings/results

Apparently phật phúc is Vietnamese for happy buddha.

Turns out he was sucked in. https://www.youtube.com/watch?v=GF3Iz7b95-8

Obviously an interesting Robotics project. Or maybe a hexapod version - some interesting Maths in getting that to run in X,Y,Z,V,W.

Homemade CNC machine with materials from a DIY, built with simple tools: https://youtu.be/mJ-TZvFpY58 Love this guy! So gloriously German! Or maybe Austrian? 'I'll be back!'

So do the Greeks funny enough and they are supposed to be lazy.

You know how the Italians have the stereotype of being laid back? Well, they work more hours than us, down south or Germany

ok

Will be half twelve before I arrive Dude.

No not yet. It's readable, though.

Well done! Have you put a nice GUI on it?

Yeah it's an unassessed assignment

Awesome! Was this a project from your course?

I also made a simple AI for it as well

No - I hadn't heard about to you told me.

Have you ever played it? It goes by a lot of names

An African board game - may be the oldest in the world.

Wow! Must see that!

Because I wrote a program that allows you to play it in java

No. I will read up about it. See you tomorrow at half eleven or so. Got you a spare phone.

Dad have you ever played Owari?

Sunday suits me. Buying you a spare phone tomorrow!

Dad if you're coming this weekend could you make it Sunday?

Thanks dude - only if you want me!

I'm fine if you come over this weekend.

I don't whether you want to see me this weekend but I am going to buy you a spare phone on Saturday! I'll give it to you when we meet up.

OK keep in touch!

Not yet I'll ask mum about it

Hello dude? You not get your phone sorted yet?

I forgot to give you granny's card - there is a cheque in it. Better send her a thank you and contact her on Facebook.

Good man!

I had a couple of beers

What did you do? Did you have a drink?

Nothing much has happened and my new phone hasn't arrived yet. My birthday went pretty well

Hello dude. What are you up to at the mo? Have you got your new phone? How did your birthday go? Please keep in touch so we know what's happening.

Back at base

OK

Train arrives at Paddington at 10.45 so should be with you about half eleven. Keep an eye out for me.

Ok

See you tomorrow then?

No I forgot to tell you that my screen broke. Mum found a spare one so she's sending me that.

I was thinking of coming on Sunday. Do you not get my texts?

Are you coming here this weekend?

Saved it! Might come to see you this weekend.

This link might be better http://onlinelibrary.wiley.com/doi/10.1002/j.2050-0416.1933.tb05473.x/pdf

Speak later. Have just found out Sean Common and Odhran Ward are staying about 100 yards from me. Small world!

I have put it up on Facebook.

Oscar Thomas Norton is the name on the article! Can it be saved?

I wasn't too sure if it was him or granda Woods

Well found!!!!

He was! He was chief brewer at Thomas Caffreys - now the Ulster Brewery.

Because I found this https://www.deepdyve.com/lp/wiley/the-institute-of-brewing-examination-ur1RzPCRjj

Hi Dad. Was Great Grandad Norton a brewer?

OK. You need to keep credit on your phone.

Still getting texts

Out of credit at the moment.

Mind you some of the answers seem to be logical enough for a robot to answer. I would say a test for autism could also be a test for a very advanced AI to a limited extent.

http://hplusmagazine.com/2014/07/28/19075/ Here is that alternative Turing Test I was talking about

OK

Apparently her phone's battery isn't working but she will call you tomorrow

Tell your mum I can't get through on her mobile.

Brilliant!

I've got one in Barcelona and the keyboard works fine

By the way do you need a cover? I won't be back next weekend but the weekend after and can get you one here. Does the keyboard work OK?

Nope the computer watching House of Cards

Brilliant! You using your Nexus now?

Yes we're all back safely.

Hi Oliver - can't get hold of your Mum. Are you all back safely?

Give this a read, Oliver. http://www.theguardian.com/higher-education-network/blog/2013/sep/16/computer-science-graduates-unemployment-bme

Very funny! Have saved it as a favourite.

Thought you might like this http://www.rinkworks.com/stupid/

Happy birthday, Roller.

http://www.youtube.com/watch?v=cx6cioPdPZQ&feature=plcp

Hi Oller Have watched this - it's very good. Bet you're glad your exams are over. Love Dad http://www.youtube.com/watch?v=BNLfNe12BKE&feature=plcp

Check this out, Oller http://www.guardian.co.uk/technology/2012/apr/01/ai-artificial-intelligence-robots-sautoy

Hello dude

http://vimeo.com/16964172 Blender and dinosaurs!

You not in bed dude?

This might be of interest, Oller http://en.wikibooks.org/wiki/Blender_3D:_Noob_to_Pro/Import_and_Render_a_SolidWorks_Model

Oliver Norton, Sarah Liscoe

How's your sister and granny? They ok?

So he only has to get up 5 mins before college!

Really? That's interesting. I'm glad you get to stay as 6th form. Olis had to move to a college. But it is closer.

But it's external to the school

I think we got Electronics this years as well

I only teach up to 100 so I never know

Yes. Something like that. Electronics.. I missed that

Three maths? You mean Maths, Pure and Mechanics and Statistics?

Chemistry, physics, biology, two English, three maths, history, geography, art, computers... I think that was all!

What did he do for his GCSEs?

Apart from art . He hates art!

Oli is waiting for his results...

I think I did, don't get the results till late August.

Ah yes. You have upper school there. Did you do well?

Nope, just did my ASes so upper sixth in September

Not do bad. Did you leave and now waiting for collage?

?

Fine, and you

Hi oli, how are you doing?

Your dad says you should be in bed

Oliver Norton, Regine Olsen, Francis Mullaly

42

What is this?

none of your fucking business

Wtf was that?

trap-a-holics

and face this

ruairi stop being a little bitch

im sittin

Caff what u doin

the cog wheel

ooooooooooooooooooo

o

ooooooooooooo

ooooooooooooooooooooo

ooooooooooooooo

you get invited to movie night

How?!

sup jessica what it do

https://www.youtube.com/watch?v=8m5CIcbytfM

jessica omg just leave

Stop being such a sad bitch

I don't know :(

ruairi barfoots a little bitch

Fuck up

tybg

shut up conrad who are you

who are all these people?

I feel alone now

its ruairi

lmao

Ruairi

;p

Fuck up you melters!

aw pads

:p xD

paddy:(

stfu xP!!

cool input jessica

*in

there's a lot of truth it that

What is this? Stfu

horse_ebooks just tweeted Im Shy

lmao

What,s Up Pete Lemme Wash Ur Legs

lil b the basedgod

because you're my sister

oops sorry im being fraped

ALL UR BASE R BELONG TO US

Why the hell am I being notified about this conversation.

no 1 cool guy wank stain

its the one and only paddy pops!!!!!

everybody!!!!

this is happening. I feel like Sarah Palin at her peak. powerful and magestic.

r u pope now

lmao

francis !!!!

lmao

wat

Hello

Pearse

Pearse

Pearse

Eva doth wish to meet me?

Ruairi Barfoot added Jessica McHugh and Jennifer McHugh.

haha catherine mcgowan

Wo added Catherine?

Basic bitches

basic

Y U DIE IT THEN

nobody cares conrad

It looks like poo

shotacon is gud

woopsies

oh fuck meant to be mchugh

ya new hair

Ruairi Barfoot added Jessica Mcshane.

uurgorgsi maisie

loll

oliver snorton on cocaine

Ruairi wha u doin

Wha

ollie nort did

anyone go balls deep lately?

shut up.

?

How many fucking people are on this shit fuck strom

lol

okay cal cool gcse results smart guy

she's fucking smart though

apparently not.

i'm smarter.

she's a smart sensible young lass

shes a very smart girl

y u leave Rose?

holla - weenus

who is she

add her again

rose lynch

yeah, halloweem. It's a jewish festival, so suck it.

Yea eamon is pretty lost

hands up if everybody agrees conrad is the funniest gal in the room

my morning jack jack

lol

Hahahahaha

conrad if you touch me i will scream

halloweem

CFinn will recieve many hugs of joyous glory tomoz.

eamon found out who his real dad is and i discovered the meaning of halloweem

hows eamon is eamon good

rofl

lost without its owner

you are literally like a little dog

hows eamon does he miss me i sure bet he does

im okay i was sick but im in tomoz

have i missed anything

good

how're you?

boom!

yah, it's been sweet

sausages

i think so!!

i think the party can officially start??

whats up steve how school been

cool chat calum thanks for making it

hey

Hello, dudes.

hes not funny at all

slit my throat

'sausages'

he just said

omg

omg

i feel more unfulfilled with life now that i have interacted in some way with him

yeah i really hate him

he's very bad

because he's hilarious

why did he

why

I'm a rice guy.

what

calum i blame you for this

you know what he just said

oh no

JESUS CHRIST

HE HAS NOT GOT TO KNOW ME

Christopher Stewart

sorry who nice guys can eat my dust My name is Conrad. I'm a friend of Calum's and I got to know Catherine through him.

he told me who the guy was but i dont know him

it was probably jd

http://gyazo.com/1b882dda7aeab083d1c583eb86945ff2

rofl rofl

say eat shit ginger

hes not the friend of me either.

say nice guys can eat my dust

say sorry who

I'm a nice guy.

hes not the friend of me

'I'm the friend of C Finn and Catherine, I was on the chat.'

i just said who's this

not yet

shall i

http://gyazo.com/40bbcc7a7a6150a7823fd38fd1e5d291

i did it today with somebody named jd armstrong

i like doing that when people add you

who's this

eva say

good

cause we ditched him

he just said "fuck you" to me on chat

conrad

rofl

he's added me

well what do you know

its just shit because he tries

yeah its called being dreadfully unfunny

hes just shit.

no

does he have a condition

lol

omg hes not good

hes not good at all

oh

hey eva

http://gyazo.com/be4510c93d6ed77a5dd550612c31b06f

hahahahahahah

lmao

http://tinychat.com/ss8b/d6c9b1/08/conrad_ulyssesmcquitty/ripcath/03-13-2013?time=1363210465046

yes

do you

i win

yeah they didnt like you

lol

i remember gerry colgan

and they were very angry

or some shit

"i love memes"

with the name

remember when i joined one of ur tiny chats

yeah

but really its just conrad

yes we were just on a tinychat with him

it was a little bit fun

conny?

who

he suckls

even steven hawking would be repulsed

hes so fucking ugly

i told him not to.

lol are you serious

two missed calls

http://tinychat.com/ripcath

and i'm like "shes dead, conrad"

hes asking me what would happen if he rang your phone

how much longer

oh my god.

i have to update flash

hurry up

and tell him hes gay or something

then you can be on cam

lmao okay

why

ill bring conrad in

make a tiny chat

what why

i do

or you do

catherine make a tiny chat

he feels bad

wow i feel bad now.

I'm sure she was man.

yeah, she was a good friend i miss her so much it hurts sometimes.

say that

say you miss me so much its painful

jesus lol

"Okay man. I'll give you a hug. And Man that's really sad"

why

he's fucking disabled

]

hahahahahah yeah

lmao

alcohol poisining? on your das birthday

swall related

how do you want to have died

jesus christ

HAHAHAHAHAHA

"really? Holy shit, when I said about going out to the graveyard I was kidding..."

lol

tell him he smells of a mistake

http://gyazo.com/03bed5c8ebf85252c84316fa0c35b282 take that catherine

lol lol

http://gyazo.com/436796c706509e9bfc1c08ed7f5c66e9

tell him something or somebody died

tell him i died

then list reasons

...of him

or that you're sick

tell him no your dog died

wants to know if im sick

i want to meet conrad

what did he want then

jesus christ

http://gyazo.com/966c36a080d4a3cf3869d9d0c30a1925

surprised its took me 2 years to say that

il be finn thanks calum

i know

BECAUSE

what a cock

its sexual

why not

because

that really made me laugh

il be finn

why not

funn

why not

yeah il be finn thanks calum

''jeez are you okay catherine?''

isnt that fucked up?

principal isnt allowed us to sing hey big spender in choir

calum holy shit guess what

so true gotdamn

isn't it just!?!?! :)

its true

yeah

eva i like that you always buy the milk not me

i dont wanna know i dont care.

its something i would tell him probably

actually no i thought i did

no i have no idea what thats about

the choir song thing

wait

probably

did you really tell calum about that

eva i like that you always buy the milk not me

calum i like that you always listen and care about my problems and tell me how you agree that my life has its difficulties for example song was pulled out of choir and you were very supportive well done your hair is also an unusual colour so well done

okay so ruairi what i like about you is that you are so generous and kind that sometimes you give too much of yourself to me, actually its becoming a problem when you through yourself at me but you do it in style

u

okay whos turn is it

lol

and hey, last but not least, eva my old pal!! i just like how u've always been there for me in my dakrest hour eva dude!!!!!!! Peace!

right eva go

i sure hope so

he'll appreciate that

lol lol

i like ruairi's unmitigated sass and perfect poise

h

O

its stil your go

ruairi

who,s next

a lover should i say

hes a real non hater

haha everybody calums a natural at this

calum that really touched my heart great start

Well, My favourite thing aout U catherine is how u dont let your disabilities get u down :P a real inspiration to us non-retards. Fuck you, you horrible gross piece of slime-shit and scum

calum goes first cause he was most eager

ew

rachel conway doesnt belong here

well you can go first them

cANT WAIT FOR THE SUPER FUN AND HEALTHY GAME

oKAY

its a super fun and healthy game

we go round and we all say something we love about one another

right heres the game!!

ayeeeeee ;P

ping

wow

anybody??

how bouts it

guys lets play a game

Fuck The Pope !!

brb :P

pope francis luv ya

Welcome To Hell

Sup, Dude.

ratty mcfatty ryt here!!

hey ruairi you fucking rat

if you go back a long time

its on fb

you said i ruined you or something

wah

i said what

ping

although ping

not the ping thing

and i laughed a lot

you actually said that like a year ago

yes, ping, that's great.

ping

oh well.

''PING!!!'' - me in your brain

ive ruined you, you dont think well anymore because catherine is in your brain 24/7

and think to yourself "wow cal is so cool!!!!!! i want to hurt myself"

you soak up my endless knowledge and catalogue it in your horrible brain

you do.

did you know nobody gives a shit about what you say

did you know that you can buy a gallon of milk on amazon for about $40 with 'no guarantee of freshness'

i dont know why it was so funny

lmao

i didnt notice she was filming me

i'd have to upload it to send it i think

is that it.

it was me drinking coffee

send me video of you

what was it

oops!!

oh

i don't have that one, you deleted it

but we were crying

i dont know why thats so funny

the video of me drinking coffee

lmao that video

do you have im only a little bit on my period

giggles!!!!!

choking lauughing at the video

i was too busy laughing

fuck i got nothing done

yeah

this was nearly a year ago

it's because we were both revising for llw

yeah lmao

i dunno i usually dont listen to things you send me anymore

the end bit with the toothpaste

thats surprising

were you never sent these calum

i couldn't make it out before

yeah, you do say that

oh wait

what is this

thats the toilet roll in my bumhole one

hahahahhahahahah

NO LMAO WHAT WAS THAT

goes

isnt that how it does

my bum is full of toothpaste

my bum is just okay

i cant believe theres toilet roll in my bum

but i just dont believer herr

my mum said she put it there

theres a toilet roll in my bumhole i dont know how it got there

GONNA WATCH IT

oh yEAH

i will play them tomorrow

of you

do you have that video

im laughing a lot it hurts

i forgot about the shaking my hand one

holy shit

my name is catherine colgan i'm shaking ma haaand

yo bitch leave it there studying cool while brushing my hair

hahaswhsr

yo bitch turn around i'm really cool and i'm really sound

i remember

lol lol

sex me babay i'm only a little bit on ma period

there's a man walking my window! he says 'hello cathreeng' LOUD HIGH PITCHED LAUGHING meh hueee a meh hueeeahhhhhhh oh i'm going crazy

yes lol

omg heya it's me catherine so here i am *laughs a lot* *cackles*... do you want me to continue?

yes

are you sure you want me to

just what you remember

so i can hear them in my head

can you type some of them

i want to hear those again

yes so much

shocking

did you laugh

oh lmao

there's toilet roll in my bumhole, my name is catherine colgan and i am really cool, etc

what recordings

im SO bored I Might Die

what Ones

again

i just listened to your recordings

but thats basically the same

well i had it an hour ago

i just had shepards pie :)

guys!!

what a horrible thought

aids

;P

...

its good we arent sexual attracted to animals

the end

but everybody died because they had no meat

and they killed all the animals

and brainwashed into having sexual desire for humans

and loads of babies were made and grew up being trained as warriors

so reluctantly they pro created

of people who wanted the humans back on top

then there was a rebellion

and soon enough there was only rape

and like the animals soon become our superiors

so suicide was really popular

but couldnt stop

like everybody knew it was wrong inside

guys imagine a world where people only wanted to have sex with animals

no

http://www.youtube.com/watch?v=LDSkvPQJ2GAhttp://www.youtube.com/watch?v=LDSkvPQJ2GAhttp://www.youtube.com/watch?v=LDSkvPQJ2GAhttp://www.youtube.com/watch?v=LDSkvPQJ2GA

brb

okay well calum wants to go out on both friday and sunday but he says its not possible how so? because he lives so far away and he doesnt want to live at mine for the weekend because it could be stressful for everybody involved on friday it will be me him and eamon having a quiet wee drink at mine and on sunday it will be st patricks day and he doesnt know which one he would prefer what would you do in his case well then, you have to ask your family on it. See if they're ok with it. I doubt it would be stressful if you guys are keeping to yourselves anyways. but then if they aren't fine with it, I'd pick st.paddy's Thank you.

http://www.youtube.com/watch?v=LDSkvPQJ2GA look

i do piss in bottles but that is not relevant

calum just told me he has run out of bottles to piss in

i hope u die

we're family here

not sure if there actually is a pickle

awk go on!!

No, I do not.

go ahead

do you wanna explain calum finn

well ariyan just gave us a hand

why

guys calum is in a pickle!

aaaaaah sweeeeeeet

all betta

ruairi

rauiri

rauiri

rauiri

haha only messing ya need education rauiri

who cares fuck school get paid yo!!

shannon collins keeps telling everybody i boffed her and now i cant sleep

i dont know if i can

i need to do essays

ya babe :((

guys do you also have this problem where you are tired but at the same time you arent??

there up ya ass!!!!!!!

mummy got 2 essays for ya

no mummy

no stop it

i hope not

the kids at his school probably treat him badly

spongebob is alive for me

i hate you

i forgot to.

why have you not showed me this

unbelievable*

this is unbelievably

sorry

jesus chrsit

m

like i havent seen the,

dont send me fucking doubleswee videos catherine

oh my god

http://www.youtube.com/watch?v=SrCXg6FFHp0

hes 12

im watching interview with doubleswee now

like theres somebody behind the camera making him do it

ooohoh

its like hes scared

his voice is shakey which makes it better

lol

eghey sexy fish

hbrr

patricks retarded

i am i cant

fucking ddr how about you end your life.

u will explode! XD

and look at how many steps there are

watch this

catherine watch that damn video

http://www.youtube.com/watch?v=Hromx6HeN_w

yep back in 1910

did she create the harlem shake

http://www.youtube.com/watch?v=mSLx_ItMJEA&list=FLDBj98IOlPH20T_Z4_-z_Iw heres something you will enjoy very much

this is so good

http://www.youtube.com/watch?v=---AtpxbkaE

and has got a gf

niall is going to gym

nall

please it hurts

niall niaall please

its the best

is this you

http://gyazo.com/50770b4b0be3d5088432f0fe26dace46

nigga

ruairi

elo DEE

actually

EL o DEE

but

yes

like melody without the m?

ELODIE

its Rocking

say it outloud

even see Gangnim style

no seriously

it was always fatima

eloldie!!!!!!!!!

STOP SAYING ELODIE

HAKIM

ELODIE

the name of the girl

omg what

what

it was always elodie

ELODIE

in your french text book

elodie

ELODIE

omg the video is still on youtube

sex cafes

randomerness!!

lmao

gettin high off weed brownies

:)

randomness

bob

cheese

farts

fingerin doots

gerry and liz fuckin

the mafia

windmills

clogs

well

didnt realise you had said netherlands

lmao

Danny McCullough, Oliver Norton

Is it on at 9.00?

it is at

i'll think about it, tell me what time

the muppets

why, what are you going to see?

wana go to the cinema tonight?

Conor Ricey Rice, Katie Galfano, Rosie Minster, Nicole McAlorum, Danny Hayes, Oliver Norton, Mark McGoran, Shane Dempsey, Michael McGrath, Niall Mccarthy, Alex Daly, Hannah Scullion, Thomas Cafolla, Deirbhile Savage, Jonny Adams, Jayne Davis, Beca Campbell, Matthew Mulligan, Jessica McClure, David Scott, Allicia Adams Rushe, Mike T. Starr, Steven Gilliland, Matthew Stewart, Tony Gilmore, Andrew Scott, Sarah Gilliland, Donal Fay, Ieuan Byers, Bronagh O'Prey, Eilís Ennis, Paul Moran, James Stewart, Catherine Jacobs Starr, Rosa O'farrell, Anna Holland, Naomi Hadley, Michael Mcdonald, Emer Gilmore, Ryan O'Halloran, Hannah Ryan, Ciara McGarrity, Megan Louise Thompson, Sarah Mc Carron

omg just picked up a new kindle for 20 quid from this discount site!! http://t.co/hpT7mkcF

Oliver Norton, Luke O'Neill

no sweat mateyo

Ok so that's a no. Thanks.

its not school

we dont

Do you know if we have to wear our uniform tomorrow?

thnx

nope just started to take notes n finishin them tomoro

Hi, did we get any maths homework?

Oliver Norton, Ross Monaghan, Thomas Cafolla, Matty Kennedy, Dylan Boyd, Dylan Stewart

Lolspam

hey, do you remember this photo? http://www.travelclub2012.com/photo-album/

100000166605340@facebook.com, 100000268464281@facebook.com, 100000380525764@facebook.com, 100000442316722@facebook.com, 100000515404733@facebook.com, 100000779674385@facebook.com, 100000817134605@facebook.com, 100000870610646@facebook.com, 100001569804546@facebook.com, 100002088593695@facebook.com, 100002112705388@facebook.com, 100003022538038@facebook.com, 100004091144351@facebook.com, 100005930497344@facebook.com, 1003970003@facebook.com, 1091382129@facebook.com, 1104291679@facebook.com, 1216121524@facebook.com, 1239587271@facebook.com, 1270907491@facebook.com, 1421842665@facebook.com, 1533604142@facebook.com, 1580658112@facebook.com, 1614090274@facebook.com, 1792114964@facebook.com

Shanter

Codecon banter

Ooh lev

lol we're joking

Lol

lol maurice I thought you were talking about codecon

What the fuck

Force you to have them thur-sun

No I'll sue for joint custody

What?? Are you honestly judging me

I'm leaving. You can keep the kids

Cheapskate motherfuk

u wasteman

Would like some of the freebies

a few

Are there any left over bags?

Agreed, I did shout out a thank you earlier but some people had left, thank you all!

Thanks everyone for giving up your time to come help today - we really appreciate this and I'm sure Freshers and parents will too!

Rip

:(

I left󾰕

We are there now

568?

cheers lads

if so send it here thanks

does anyone have a photo of just now

Yay

We're coming back

Maurice and I have the sandwiches

What's the situation on swag bags?

It will be worth it haha

Good progress on baguette trays

Sorry for keeping you guys waiting

Stop pissing around

Fuck me

Think they got distracted by bev

James Long?

Where are the bags

Yay we need da bags

PAGING ALESSANDRO

Food is coming soon

Alright

341/2

Where should I be going?

Wait you're in 568?

no copy

levlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlevlev

Lev come 568

341/2

We're packing in 341

Not common room

Pack

To pick

Should I go common room now?

eta on food?

/=

Almost != Almost

Sorry, food is still not ready

Should still be open

Food is almost ready

Thanks

Rightio

Indeed

146

Yep

We need to get the swag...

Umm guys

We may have more in the cupboard

Go for it

There are 9 cans and 12 mini orange juices

I'll go and get them

There may also be some snacks and stuff left

Feel free to grab them

Es gibt drinks in common room

L'Omelette espagnol

il y a un piscine dans la ville

ou est la biblioteque

Omelette du fromage

J'adore Le fromage

Le baguette

Ah oui

We'll be there shortly!

I'm waiting in 342. See you soon

Getting a selection of baguettes

We're almost there now

Sandwich shop

Gonna be 10 mins late probs

Yo I'd happily have some food

Which shop you heading to?

Gonna be late

Bag packing

yes

Something is happening today?

In time?

we're going to sandwich shop

can't order in time

can't do subway platters

Love that

Probably should get at least one veggie platter as well

Yeah me and Lev always get that one

thanks John, will get that

found the email with password

Spicy Italian on classic sounds nice. As an aside how come the meat feast platter features veggie delite

login details not on page

If you need a card to pay with let me know

Cheers Maurice

ok

I'm getting back now

I'll do it

http://subway.co.uk/menu/subway-platters.aspx

(I made an account a few days ago)

I will come direct instead of via jcr if there is food

Password should be in the document

I'm at the gym

What food is available? Is there a. Menu

alessandro if you're free?

could someone order platters?

Will be there in 10

They'll take 45 mins at least I reckon

LOL order the platters stat

yes

Yes

Yes

Looool

Yes

This is like the Yes chat all over

yes

Yes

Sixte you coming?m

Yes

Yes

Should we get Subway platters?

What do you offer

I want

That's great

if you want food, please say what you want

we'll buy you lunch if you want

booked

12 - 3

341/342

About to cycle over

maurice are you near

ok lads

Lol

Tbd

The

What room at 1230?

Lol

Lol

Wait...

Chirpse

haha

Lol 󾌴󾌴󾌴

Paul Balaji Ale Bonardi

PBAB

PPAP 󾮟🏻󾮟🏻󾮟🏻

PPAP song ?

https://www.youtube.com/watch?v=_yXGxL7dfSg

we need a work song for tomorrow

Lol

Make it the docsoc chant

Thanks fam

Give us a beat

Paul Balaji, your Time to mix has come

5/7

Sick mixtape

Damn

Yes Vim

Wanna listen to my mixtape?

Lol Sickenzo

Sickenzo xD

RIP Sickcenzo

Rip

I'm sick soz

Chirpse?

I love filling my sacks

Room tbd

Hello guys, tomorrow 12.30 we are meeting to pack the fresher bags :)

100000515404733@facebook.com, 100001866181209@facebook.com, 100006776130896@facebook.com, 1541047107@facebook.com, 650331469@facebook.com, 660934412@facebook.com

Haha cool

Nothing tomorrow

well if you're not going in then have a chill day

idk what to do

not at 9am haha

are you planning on going in anyway?

you could ask on Piazza or email her but I don't think you'd get a reply before tomorrow :P

hmmm idk

I'm not sure

It says lab But since we haven't done anything I assume there's nothing happening?

is there machine learning tmr? or is it lab?

We done did it

Bruce and David, sign

Great thanks

ASAP by default you do the coursework

and when do u decide?

And the exam is still 1:40 regardless

And also the coursework is in matlab

And if you do the coursework then the exam is only 2 questions instead of 3

I'm not sure if it's recorded The assessment is it's 33% coursework basically if you choose to do the coursework which is optional

im thinking of taking it

is it recorded? hows the assesment like?

did u go to the lecture today?

But totally still need a group for machine learning

For the business and management thing

Just gonna throw that in there

Also David and I have found a group to join

yes

to send an email

is it so hard

mate

that was ages ago

three weeks? minimum

nah fuck that when did you ask though like how long are we talking

xD

and type an email

use your hands

im not asking again

When did you ask tho & nah you're group leader They said to contact the supervisor through the group leader ;)

your turn to go after him

i asked will, he said wait

they said ask will

i already asked ms

lol

yes, you can ask

But like for real Can someone ask

GIVE US FREE STUFF

Did anyone find out about when we get our Microsoft things

not sure whether to do logic-based learning, distributed algorithms or graphics

nah, I might download the slides or something but I've heard it's a lot of work or it's hard or something it's interesting but I wanna do Machine learning and the security one so I only have 1 choice free lol

Aww cool the border turns yellow when you submit on cate

apart from oliver

are any of you taking graphics

nope

are any of you taking systems verification

fair enough

need to eat before i pass out

ok will do in 25 min or so

Yeah just send it to them then

Forgot about that.

and were looking at our code throughout term

damn i feel dumb

like they set up our repository and stuff

didn't they have access to our source code anyway tho

Because we might be able to give them the report but not the source code or whatever.

would be good to double check

ah i think its fine but i can check if you want

If it's ok to send the whole thing to them or is it fine?

double check what exactly

Aren't we gonna double check with Will first?

ok?

in slack

i will link zip to AA

cool

uploaded to git

yes

So report and presentation are in there

to change stuff if needed

we have until wed 18:00

aight it's submitted

Ah

Did we put comments and stuff?

unless anyone wants to write any comments (there's some already in the clearpath stuff) it's done

is there anything else to be done to the code

last chance

right

https://docs.google.com/presentation/d/10Yw4rmW5zx41Tl2DbfBBiQI8mBTnlxn6vOnRuQkGsb0/edit#slide=id.p

slides

https://www.overleaf.com/7599531csvgzqmjvdcw

yes why not

To report and presentation

what links

not bad

license and everything

Shall I resend the links?

aight

Put in the report and presentation and it's don

is the archive ready?

made the group for the code submission

right

cool

after systems ver lecture

Although I haven't put the slides + report.

Oh Paul I've merged everything into master if you want to submit everything.

Tbf ^

it can be understand indeed

m8 I don't understand English and I'm English

yeah there's no gif of it which makes me sad but youtube to the rescue

understood*

English is a hard language but can be understand through a tough thorough thought though

thought*

yeah, I though of that one and tried to find a gif, but failed

https://www.youtube.com/watch?v=uAtLV26wnCE&feature=youtu.be&t=5

ahah, well that discussion with that group was going before you suggested this group :)

but that's not 100%

So I think me in Polly are in some other group

Yeah so Misha talk to Polly, sort out what youre doing :P

aight lets wait for them to respond just in case xD

i c

Twas what I was aiming for

What*

That Wikipedia definition

Middle names omitted because I have no clue whag they are

Polena Lilyanova is the girlfriend of Mikhail Bobkov and a third year computing student

who the f is polly

Well I've known pikachu for longer, true

pikachu overrules Polly

pikachu wants you to join us

Well it depends on what Polly chooses as well

Hahahha

Join us

well pick ours and then you know for sure :P

Yeah, there were talks about some other groups so I am not sure where i am now

well we don't know the group size do we on piazza he said like less than 6 or 7 I think which is hardly a solid number and he's meant to be teaching us management ironic

If we get misha and polly then that would be 6

4 :)

Just have to get two more

I'm fine with this group

Well that makes 3 of us at least...

I'm down for it though I need a group so sure

that's what I mean though is everyone up for it lol like MIsha hasn't replied and I think Polly said she might be in a group with him and some others Oliver hasn't said anything so idk what he's doing

do i have to be nominal group leader or is someone gonna invie on cate

forming a group

also the deadline for this group formation is tuesday so.... what are we doing?

that site does the hottest sauces I've ever tasted I have an extreme tolerance for hot sauce and they kill me lol

http://hotsauceemporium.co.uk/shop/hot-chilli-sauces/psycho-juice/

I've got a few ghost chili sauces that are pretty brutal

who wants one

http://extremefood.com/Ultra-Death-Sauce/

Hopefully not economics I hate economics

The management and business thing

ye?

its for the economy thing

ay

Well Group of 5 then

I already have a group sorry 😐

I mean it's due tuesday so if nobody has any ideas then we could stick with this group at least then we'd have a group..

So we could do basically the same group lol Hopefully they'll give us more information.. Something solid

https://www.youtube.com/watch?v=HgQEuPw942c

they dont fucking stop do they

ayy

tuesday

whens deadline?

Oh, It might only be straight computing then

i gotta do it too and im beng ._.

I think MEng (me included) have to do it

apparently it's groups of no more than 6 or 7 according to the piazza

who here has to do this management and business for computing nonsense

Alright

can someone send all 3 demos

Should've asked a question lol

I didn't quite understand what they meant when they mentioned "several iterations"

That's a shame Did they say that was all they could handle or was that just the biggest they had Would be cool to compare 100 for both

They had 50 drones but only 4 or 5 were interacting at a time

Did they just do it with normal game controls I assume Like keyboard stuff Did they have a big demo like our 100 one?

Pretty good. Their last bit with real time control of drones was cool

How was the other group's presentation for any who went?

Tbh it's on us for forgetting to tell you 5 of us You woulds thought someone would remember lol

labs

Ill be there in about 13 mins 😬 sorry i didn't know we were practicing

Just arrived in labs

so much fun to have the flu

Ok

We were gonna put that in the chat

Oh did anyone tell Mery that we're meeting at 9:30 to practice

cool, it's in that one

yeah just loaded it up and it isn't in there

oh shit it's not there

cheers

K

preferably in the test file already

can you send me the 1000 drones one

yep

is it all set up so it's demo1, demo2, demo3

yep

ok, safe to delete then

yh if we're demoing that we don't need the pictures

is slide 12 now superfluous?

i'll just explain it briefly

ok

yes

so you're sure slide 8 stays as is yeah oli?

okay

I think it's fine the idea is, or at least what AA told us at the beginning, that when red spheres touch we send commands the red spheres intersect but we're sending commands so all good

ok

just informing you that they do cross

im fine with using that version of 100

lol

explain that to them if they see those two paths

The whole point of the clearpath algorithm was so that they WOULDN'T get too close.

plus for demoing purposes at least it's fine

nah it's fine if there was an issue we'd have a test failing

xD

end of story my friend

if red spheres intersect theyre too close

like they're going in completely different directions

but the thing is they're not actually too close

they get too close

yep that's the problem with 30 10

that's not the angle

um

i've changed demo1 to that

it does

imo that looks better than the one we're currently demoing

like that one with 100

if you're looking for better files to use though, I'm getting smoother ones with 30 and 10

yeah you're gonna have to signal when you want something or if you want the demo to be finished btw coz idk what y'all are thinking

we're not playing the whole 100 drones thouugh

yeah search of 30 and drone radius of 10

btw if we go over by 2-3 mins by making the 100 drones one slightly longer its fine

and the constants I think are 10 and 30

well I uncommented my multi=threading stuff that's in there which probably isn't the most efficient but meh and made the server more thread-safe

did you change anything?

it passed

no weird loops

it's beautiful

look at that

but ignore that

plus shit like if your computer is running other stuff in the background it'll be slower

okay

and also general millisecond time delays

Because they wait for the response of the request

and why, on the test harness?

Any sort of requests and test harness

what part of our system includes delays?

yeah that's probably the best way to put it

When actually running the whole system it's not

The algorithms are deterministic.

but I guess the answer is basically no lol

because of the server and test harness side sending off at possibly different times each time and then there's any lags due to actually sending it over the web so no but then the maths if you did it by hand or whatever would give you the same result every time

In theory it is but because of time delays it isn't

Exactly

yes

but in theory

well no

if the question "is the system deterministic" is asked, what's the best reply?

no one cares about the approach

yes, that's what I meant

to watch them evading

why slower than 1x? that'll take a while I guess slow it down once they reach the middle-ish

yeah Oliver it's the file I just sent most recently just smoother and doesn't have the sudden jump in it lol where they suddenly accelerate and then go back to normal

You can just explain that it looks like that from that angle then rotate around.

and let me talk through it while you spin around whenever it looks like an intersection from the front

make sure you put it on lower than 1x speed tomorrow as it's last demo

same, I couldn't see any obvious real big intersections or anything

it passes

yeah yeah

does it pass?

but someone should double check

didnt see a problem

i just stepped through it carefully from several angles

so that one there?

haven't checked if any red spheres intersect or anything but fuck that

it does

looks nicer

that one doesn't idk if it's better

the one you sent has some random loops in it

that one is pretty smooth

the webpage is fortunedotcom

im stupid

random economy magazine

what is fortunedot anyway?

no fly zone and manned aviation

i've got that one too.

yeah

3 demos total then

cool

fortunedot

what was the source on the graph again?

this is the best 8 drones one i got

I'll do it later I'm playing dark souls 3 now

pull again

as in I didn't push

I pulled the changes so I'm up to date with whatever was last pushed

probably that i haven't pushed the latest

Don't know

Oh Why does it say features now lol

the top bit where it says features should say telemetry

doesn't work for me

k

I'll just run it again and restart visualiser etc to make sure

send me one of the new logs

like I typed app3 into it and that one worked

it works for older ones

obviously lol

is the visualiser running?

nope no errors

visual studio gives errors

weird. Did you check the format?

but none work

and copied them across

like I just ran all the tests

the visualiser isn't working for new log files for me

I'll just explain it

Yeah the screenshot of the test harness doesn't really work.

better

Ok how's the graph now?

You coming in Misha To find out

Come in and find out

I'm not living 1 min away anymore, so I would rather not go there to discover that I don't actually need to do anything

Come in and find out

is that just presentation practice or are we doing something else as well?

ok ill come in now then

I can come in now if you want

Well David and I are in labs now

right ok

is anyone else presenting?

what about everyone else?

i can come in then

ill be in labs at 10 past

What time is good for everyone

Hmmm we should practice at least once more

100 drones test passed boy!

dunno yet I'll probably be in labs from around 3.

What time are we meeting?

So you can return a tuple. One part with the result of checkCollisions and the other for done.

also to check if it's done

to figure out if there's a failure.

Then you can AND them together.

for checkcollisions

what result do you want though?

seems like

So each thread produces a result?

Is there a way to do Future in C#?

not sure which approach is the best, have to to toy around with them Also got the issue of synchronisation issues everywhere but I handle those as they fuck up

like updatelocation, checkCollisions all that stuff as 1 job? would also need to do the same with manned aviation or leave that in a loop but could do

Ok how about each job just be looping so there isn't a "main loop" but while loops for each drone?

I was also thinking of making it so you can check when a job is done if you come to the start again and the job isn't done, don't create another one

ok

start() ? all it does it queue jobs, the threads already exist

Why not build the threads outside of the loop and then call start() at the start of every loop?

yeah that's the problem with sleeping for a second, but they only need to send telemetry every second but then the test harness becomes less precise which was kinda why I left it doing it every loop iteration means they get updated regularly at least

?

So they only update their location every second.

or I guess if we don't want it to sleep for a full second we can make it do less or no sleep and just while(true) updateLocation

I was thinking of making it so we assign the jobs before the loop but the jobs basically just update the location sleep for a secon then update again

yeah that was my thoughts

Is there some way to move it outside of that loop. Because recreating the threads for every loop is pretty wasteful.

the one that updates all drones updates all manned aviation then checks collisions etc

yeah it's inside the main loop

Is all this inside another loop?

each loop queue up updating the drone's location one thread per drone

a thread for each drone for each loop?

Wait, do you create a thread for each loop?

I was planning to change it so we don't assign jobs

it's probably the overhead of creating threads and queueing them and stuff

That's weird

it was actually slower than doing it without threads for 10 drones

hmm that's a possibility I had it waiting for all the drone updates to finish before going into the next iteration of the loop but that caused it to slowdown quite a lot

That's exactly it

unless you're saying two different threads of the same drone got the commands and tried to do something

that was test harness side tho

Two telemetry messages by the same drone are being processed concurrently

Ok I think I know what's going on.

genuinely confused by that one

that is weird

targetVelocity and c are both not null and this is in the drone class so there shouldn't be any synch issues?

that's a new one

wtf

where dronerepository is the name of the drone repository

just do lock (dronerepository) { }

yep that's it

is there any? I guess need to add locks around when it adds and removes from drone repository? seems like something that should be handled for us

Synchronisation on drone repository?

???

that's weird how would a null drone get added

So a null drone must have been added to DroneRepository

sorry I see now

is d null then?

with d as drone as an input

yes

a predicate

drone isn't null oh so you're giving it a function to evaluate?

check if drone is null.

it's a lambda

d is just a parameter

but wtf is d what is this doing

what's this about obviously d being null is the issue

I think if we'd made it all threadsafe from the beginning we'd have a lot fewer issues right now lol

goddamned synchronisation issues

yep

with the log files?

Synchronisation issues

nope.

did the 100 drones one work?

is that any different? looks the same as before to me

Here's something interesting for two drones. Commands would exist for 2 seconds but I changed it back to 1.

I got the new logging system working. Going to test with 100 drones

the test is passing most of the time concurrency issues are a bitch

oh wait no it isn't literally the block above that checks it and if it's null it should add it

which is fair enough because it isn't in the drones repository apparently

this time round anyway

yeah d is null

dunno, I'm trying to get it again I'm just running the 10 drones at once test seeing what problems I get

is it?

a guess

that a guess or have you managed to reproduce it?

d is null

as per usual the obvious stuff isn't null

nah

is location null?

but as you can see from the screenshot up yonder ^ other things seem to die too

& yeah I added a lock on the bits that change

Already got that sorted for logging

so I think we need to make our server threadsafe

kdtree is global

you shouldn't have issues unless you use global classes and stuff

yeah a guy says there

oh

I think it is. http://stackoverflow.com/questions/20230093/must-i-add-explicit-threading-to-my-web-api-rest-methods

is the server somehow also multithreaded? I woulda thought even if the test harness was sending off requests faster, the server would just queue them and deal with them one by one

im ded

everything is going wrong now

reading is obviously fine.

Adding a lock on the kdtree for writing should work.

they'll be random and maybe loads at the same time

yh which we should definitely fix considering with 1000 drones they're not gonna come in 1 by 1 lol in a nice order and the same order every time

so there's probably a synchronisation issue on the server end.

which is definitely working, went from sending telemetry from drones 1-10 from doing it in different orders

yeah only seems to be once I introduce the sending telemetry in parallel

I removed the multi-threading and am running it a few times but I never seem to get that error until it does the updates in parallel

Yeah, that wouldn't make much sense otherwise non-parallel test harness wouldn't work either.

the kdtree is server side

Do you have a lock on the kdtree?

that's the idea anyway

it queues up all the drones location updates and then just does them in parallel

ok

no no it sends off the telemetry concurrently

Wait, are you running multiple tests concurrently?

oh

but multi-threading in the test harness shouldn't cause the server to break unless it just can't handle the load

I think

although it does only happen when I add in the multi-threading in the test harness

so god knows

I get the feeling it's happening IN the kdtree stuff which is something we didn't write, right?

Maybe it was already deleted or didn't exist in the first place

Not handled in user code.

Nope kdtree isn't null

lol I restarted vs for test harness and server and you know what it did work that time

on kdtree I mean

That would make sense since you're calling the method on null

is kdtree null?

maybe

will that make a difference?

Have you tried restarting VS?

but it isn't null

you'd think that

that's weord

But it shouldn't be because of the try

oldPoint is obviously null

it's removing something dodgy? or it's not there?

what could that be caused by

That could work too.

fair I guess another option would be to just log it to the file and then when a request for the file is made, just do 1 pass and turn it into json

Thats why why I'm going to write to two separate files.

It's a json file of two different objects

lol what, really, why do we do that

Which is obviously terrible

Write as we go, but because of the structure of the file we read in the entire file every time.

how does it actually work atm well how does it, barely function

do we write as we go, or store it all and then write as one big clump

Writing to two files, one for telemetry and one for no fly zones and merging them.

I'm looking at rewriting how the logging system works today anyway

Yes

does it just randomly crash in the middle of some random test?

Maybe a memory issue?

Ive tried running the tests twice and it happened both times

I've only had that once.

visual studio

and then it restarts

nope just "Visual studio has stopped working, looking for a solution"

Hmm. Crashed by an exception?

yep

Did you pull the latest?

That might be it

not sure I'm just running all of them

Even without the big test?

they run for a while but then visual studio crashes for the server

the server keeps crashing visual studio when I try to run the tests

cool, we can meet when he's free then

I think so

ight, we can do that later then I guess if it isn't done what time are we meeting, Paul said he was free from 3 was it?

No, not yet

Don't know. Haven't checked yet

have the improvements been made to the slides?

In regards to today, I think we're going to practice the presentation a couple more times later

sweet cheers

ah yeah bool overwrite

By default copy prevents over writing

You just add true as a second parameter

Oh that's an easy fix.

oh or do I need to delete it first

I take it the storage stuff isn't fixed either?

it's weird coz if you check the location and stuff, nothing is null but yeah

Don't know. Maybe some of the data was lost? I've rarely seen that happen.

why that happen

what's the plan

so what we doing today

see what it does

maybe add another drone in opposite direction, 10 metres upwards

yep, I think Paul said he'd be in by 3 that might change by tomorrow

Are we going to practice the presentation tomorrow? I don't have an interview :)

Still not sure if it's demo worthy But it's an improvement

I'm going to make nfz more visible

O nice That does look good

k

Post a screenshot of the visualiser?

2 no fly zones

Check this shit out

that's cool

and it would run first, second and third tests

so you can type "0, 1, 2"

the numbers

how do you choose tests? can you just type the test name I've forgotten most of the test harness stuff tbh

The part of the test harness where you don't have to run all of them

Not that part

the selecting tests thing that we haven't implemented? like being able to browse and select test I mean

second batch probably soon

lol

and thye've sold out

IC hack tickets are on sale if you guys didn't already know

That selecting tests thing is really useful.

Fair enough

i saw it

in somewhere

not in paper

there is in paper

BTW there's no trademark for the ClearPath algorithm

Be a bit more explicit that ClearPath isn't ours.

Have K-d and collision detection together

Spread clearpath on more slides

In terms of the slides:

as Obstacles

RVO2-CS

https://github.com/snape

Ok noflyzones already exist in the 2D library

that was literally, as the saying goes, about as useful as nipples on men

ayy lmao

where is the c++ and c# code

need to check how its done...

c++ one is structured differently for efficiency

in the c# library they use obstacle class

so what's the deal with no-fly zones? what's the file Paul was saying about, Obstacles or something is it translated or what's the deal

Oh ok

No, we're done.

Still practicing?

are we gonna find a room or are you all still on the sofas, or common room ?

Ill be there in 10 too

yh looks fine to me where you guys at I can come in if you wanna practice it

changing

fair enuff

mm personally I think it looks weird I'd say definition then sub-bullet points would be a better format which shows it's the main definition but the grey looks odd imo

thought it was useful

definition in black, extra info in grey

why?

oh it looks odd

that and the kdtree are intended

the VO one

the clearpath slide still has grey bullet points on it

wrong file

done

rather than collision algorithms which would imply they're trying to collide lol

it should probably say collision-avoidance

on the objectives slide

bruce, proofread

presentation os done

right

that is pretty smooth

ah you found it

so it'll be one of the app ones and you just want it smoother

the drone vs drone scenario were they both curve away and return to their path really smoothly

we had loads of those

the nice one which looks like it was drawn with a curve ruler xD

nono

need the one with the curve that we had

you want it zoomed out?

that its way too close

what's wrong with the one we have

fair enuff

smooth clearpath log?

prefer not to, I really hate public speaking plus I don't really know much about anything

same angle as original

and ss it

of drone vs drone

find me a smooth clearpath log

there is something to do

idk just asking

?

so youre not saying anything

kinda hard to practice on my own lmfao

changing slides according to the pace the person is speaking at

tell me how long it takes

practice whatever your part is

http://prntscr.com/du8ix4

is there anything to do?

ah, well solves that lol

correction labs are not open

cool no point coming in then I can't do anything on the project on those computers really and presentation is mostly done

lab is open

got one left

just did genetic algorithms slide

ill check

I did some of the genetic algorithms slide but presentation looks pretty done now so no point coming in anyway I guess

sweet is labs open yet do you know?

u instead of 1/2u

yes

we tell the drone to fuck off instead of moving the plane or w.e

ah alright figured it would be except we give them high priority right

it's near the same as drones

It's easy to implement in clearpath

but you said it doesn't work with manned aviation? or are you running it on the old server version

no I'm looking at the 100 one with manned aviation

anything else you need? that doesn't involve changing tests

cool coz it was a nightmare getting that should be easy, but everything was going wrong

that's exactly the problem we had

yes

looks good to me yes

weird paths and they get too close, that what we wanted?

finally managed to get the files for the server_old

on objectives slide

now theyre in black

oh I know the tests[7] wasn't intentional I meant the bullet points being different colours it looks weird but if you're still working on them then fair

afaik

that only happens on VO and KD tree slides

It was not intentional in that I didn't mean to commit it

which im doing

apart from objectives slides

yes

is that intentional

and most are in black

oh btw why are some bullet points in a grey kinda colour

for some reason

oh lmao when it does runTest it's doing runTest(tests[7])

but the file is definitely the no fly zone one

but then weirdly it does say "Test Drones approaching head on passed"

tests.json has the right tests in it

why's it just running that test?

I'm on server_old

Unless you go to server_old

We don't have it check for manned aviation yet

alternatively stick a plane in 100 drones

simple, 1 drone 1 manned

someone make a test with it

yes

in the visualiser you mean?

thats fine, just want to show how we represent it

do we have manned aviation doing anything in any of our tests? the 2 drones head on has it but it's like miles out lol

Cant run visualizer on my comp

Yes pls

mery wants for presentation

somebody make one with manned aviation and ss it

that's the no fly zone if we wanted that

hm

it was godly

yeah my one was literally a perfect circle

that I had

that's the roundabout

http://prntscr.com/du7myz

choosing best

i suggest running the intrresting ones 5 times each

I ran the old one but for some reason the only test it runs in the no fly zone in the middle of path so I have one for that..

did we need an example of no fly zones?

I've got one for 8 drones here

I've been replacing them as I generate them

nah it didn't work for a couple of the others

we should have the log though

did you keep it

it was amazing

they all just went anti-clockwise in a circle until they got to the right place then carried on

I can't remember what values it was I had one where the 4 drones going head on worked so beautifully it was like a roundabout

choose the ones where clearpath looks amazing

I think we wanna use the same test, for proper comparison the 2 headon is good, simple and shows the differences just need to figure out which branch has the old algorithm and stuff but is all working

choose another test if you want

I took it with no intention of putting it in

or maybe find a file earlier on in the chat and just rerun it through the visualiser

I guess we can just checkout to an old branch and rerun it

interesting choice

it was, I took it

other one is fine

yh the original algo image looks like it was taken on a camera

and I want max marks, idk about yu

either we follow their guidelines or we get marked down, even if its a tiny bit

same thing intake of information

were not there to teach them, were there to show them a product

OS guy was bland and monotone still taught me OS

yeah that stuff is bullshit you get information from their body language yeah I totally learn all my lectures from the lecturer's body language yknow as opposed to the stuff they're saying which is technical

look at the quality of the other slides and compare

at least the left one

take better pictures, for a start

whether we like it or not

they will be looking not only at if we know our stuff but i our slides are decent, we present well, the 55% body language is there, etc

what's todo on the clearpath vs original algorithms slide? pictures are pretty self-explanatory

and what do you think theyll be noticing in real presentation mate

blah blah

your font is too small

nah they were like "where should you put the demo?" this is what I think what do you think? what does that person think

https://docs.google.com/presentation/d/10Yw4rmW5zx41Tl2DbfBBiQI8mBTnlxn6vOnRuQkGsb0/edit?usp=sharing

actually I think the feedback is decent

it's all opinion

shitty feedback

come to uni at 13:30

bruce

We get feedback today so we know what to fix

no difference

this one is just in a room

tbh we're gonna practice it on thursday anyway, or friday

you can work on presentation by filling in the slide on genetic algorithms

its not as if were gonna spend 5 hours practising for the practise presentation

do you prefer zero practice over 1 twenty minute practice?

is there anything useful to do atm since we can't use labs don't wanna work on the presentation from here in case it clashes with changes you're making

We just have someone who's normally speaking to run the demo

because he'll have no practice and we'll be doing that part with him

meaning the timing will be wrong anyway

yh but is the thing even done and we'll be practicing without David

We can meet at 13:30 and practice once

but we need to have the timing at least partially right

if you wanna get there at 2, do that

just my opinion

i just wanna finish and leave

do you wanna be destroyed by catarina for running the presentation for 10 minutes more? xD

Are we doing that? Seems kinda pointless

omw in in 10 mins

feeling fucking terrible, fell asleep

come before so we can do a test run for timing

Well you're still coming to the presentation at 2, right?

Laptop

No point me coming in if labs is closed is there Given I dont have my laptoo

Dunno

Oh When do they reopen?

Labs closed for test

I'm just on the sofas on 3rd floor

I'll be in 15 min

Who's in labs atm?

welp ill be in at 11

If we have to we can just the original algorithm.

No fly zones would be super

Logging is def an issue Even if we get that test to a point where it'll work we can't currently get the log lol so yeah

Though I might look at logging again.

*that

just hat

Probably is.

But that just means changing the test i think

Are we gonna try and work on the project more Like what can we do in the remaining time that we can demo and is cool? The 100 drones one where they actually reach their destination would be nice

4 slides left

yah ill be in around 10

Who's going in tomorrow then? Is it everyone? I can't promise 10 Depends when I feel awake enough to deal with this But I can come in

Honestly, we're better than toilet humour

Figured it out. Apparently you have to flush afterwards (no toilet jokes)

as in no exceptions. The file is just empty

Oh btw writing locally didn't work. I don't know why though.

what time's your interview David?

damn david back at it again with the zero contribution

you wanna throw me out dont you

what, the lack of art sense?

That's why we're at Imperial

we have a presentation thing at 2

jeez

no sense of art in this group

tetris picture represents components that can be added

Idk what time it finishes, if you're still going to be in after it finishes I can join you

Yeah I have an interview tomorrow so I won't be in

Well, I'm coming in at 10 to go over presentation and other things

damn David back at it again with the interview

David has an interview, right?

so what's going on tomorrow

lmao the server is running at 1.3GB

what's up with that tetris picture lmao

I'll give you the 1000 second one asap

Ok. I'll stick that in.

can we not just get it to save the file locally like if you're running the server on your laptop fuck going and putting on Azure and then getting it again why not just make a file and leave it then we can just do it manually

Ok so I can't seem to fix the log. I just reverted it back to it working with everything other than huge examples (100 drones for >100 seconds)

Yeah I can do demos

can practice timings on thursday I guess

I'll do slides

so as not to do it while were talking

and whoever is not talking does slides

mery does visualiser

somebody takes care of all demos (david?)

i do clearpath

oliver does original algorithm

so for now we have

I thought I did, but there were still issues

ah ok I thought you had it working again which was why I was slightly confused there

Like I said, I'm still trying to fix the log

so do you know what the issue is? or is it just happening to me

im working on the log now, so I'll get back to you

It didn't thrown an exception, unfortunately

The last time it caused the whole thing to take up 3GB of RAM.

that might have been it.

tried 3 times, two are that error, one was a separate issue but it got further than that

Oh

It worked once this is with the new test file you sent me

Has it worked before?

oh well this is the second time I've gotten it

i've never seen that happen

No, that was a timeout

was that the thing your stringbuilder change fixed?

but we need to either account for that or make them finish further apart

yeah can't remember why, have seen it happen tho might have been with a massive search radius though in this case from what I can see yeah it's drones finishing too close

They literally shouldn't be doing anything than straight lines unless they're close to another drone.

Can't explain that.

sometimes though they're not actually close to any other drones but do weird shit

Ok, what we could so is if a drone is done (in the test harness), we remove it from the world.

It's because they're too close to other drones at their locations.

some of them do some weird shit but idk where they're trying to go so can't judge

CHAOS!

If you guys haven't filled out WebPA, you've got an hour

depending on time

yeah so I think we don't really need one can be spoken, like after the demo just like, "we found that..." blah blah any questions?

last bit being the clearpath algoruthm

Well, that's where it lines up anyway

and then go look it's better

I think we kinda conclude when we show our original and then the clearpath

We should have a bit of a conclusion, though

exactly

and can probably cut the one about "what we learned"

will need to shorten some bits

yes

easily fill the time

I think we have enough to talk about

oh right not needed then

Because we though the presentation was a summary of the report

that was a filler slide that we might have needed

what's the "approaching it differently" one?

after that should be fine

need objectives slide done

then moving on to algorithm

doing two slides per component

still looking pretty noice

how's the presentation going? last I looked there's some nice diagrams there

Just heading now. Should be there in ~20

Cool. I'll do that ASAP ^

somebody needs to render an old log with the drone avoiding a no-fly-zone

weird and funny tho

prolly is, the 2 sets of 2 drones is fine

I'll have a look once I reach uni

that should be the log

Obviously there's something wrong with the log data

That must be a error visualiser

it passes they go up to the point they turn off fly off and come back in like an instant and then carry on normal speed

Wut

um

what the flying fuck

cause my flatmate is too annoying

after interview i came right back

ill be here till 10 or so like yesterday

I should make it to college by quarter to 4 if anyone is still there

Right sorry about that. Was for much longer than I thought I would be.

the usefulness and stuff should go before all that

place

yeah that's what I meant that'd be a good palce

i think we should start talking about the components after the structure slide i just did

then we can leave the test harness there and just talk about algorithms for the rest

maybe it should go after the diagram of how it all fits together?

lots of talk about algorithms a demo then like btw slide on test harness

and will be put right eventually xD

everything is out of place

the Test harness slide seems a bit out of place

To watch others?

you can do what you want

well im definitely going

So we don't have to do it tmr

what bruce said, again xD

Is it compulsory?

so I don't think so otherwise we'll be rushing and it'll be shit

we should yknow finish it and practice it a couple times before doing a rehearsal

we don't even have the presentation atm

yh but I doubt it

yes tomorrow wednesday

what bruce said ^

Do we have presentation rehersals?

I think Paul is in uni? I'm like a minute away coz I live close but I guess we don't need to be in labs to work on it it's just easier

Yeah i know, i can come in now and work on it if its useful

but idk if you go any bigger you seem to get weird paths

& they go too close though maybe the red spheres do intersect

it's on google slides we were waiting for Oliver since he has the most ideas about what's going on lol plus then it's 4/6 at least but idk he hasn't been on yet

google slides

im already working on it

wait fucking finally

it's beautiful

To work on the presentation

When are we meeting ?

anybody have the slides from AA?

ayy sorry for that mates xD

oh that's helpful at least lol a test creator would be 👌 but too late now

well at least it gives you a lat lng

because that would be beautiful

does that shit work for creating tests?

thank goodness for the click listener in esium xD

same just copy the old format and then I guess you have to work out the long and lat stuff

I'll try but I have zero clue how to write them

idk the test might be fine, it worked the other times who knows yeah that seems good go for it

also I thought we could write a test with 8 drones at the corners of a cube trying to get to antipodes

see what happens

change the position of one of them to the left by 1 metre

maybe try modifying the test file

I have no clue

it did time out not fail weird path but technically no collisions

yeah lol I'm running the right log I deleted them all and I move them across every run so defo right one sometimes it might just go funny I guess

i know app52 wa decent

somebody should make a list of the tests that actually look good

you're sure you're running the right log?

weird that the drones going in weird circles aren't gonna collide or whatever but are just doing weird shit

which is weird but that's what they were when I pulled it

still search of 100 and drone radius of 20

haven't touch them

diff constants?

works mostly but that time it timed out some weird paths

wtf is up with that 10 drones flying at once

going to bank, food, then lab with you guys

right im at ic

Oh yeah We all gotta do that

Yay, I've done webpa

good thing you got til friday to recuperate then

im sick 😂 gonna be fun talking in presentation with half a voice

I tried using waypoints to determine desired velocity. Did not make a difference.

That would be bette.

so are we meeting tomorrow at 1? or do you not know when exactly you'll finish Oliver So shall we say wait for you to say you're finished then plan when to meet?

So if there was an element of randomness you could put in for choosing the velocity it might work better

I think what happens if that the only applicable velocity is backwards, then they start moving back

im tempted to see if there library took a shortcut somewhere and is searching a cube or something

It's with a small search radius

i know

that headon one is fucked

Interesting, with a lot of these values for smaller search radii, the zig zag

That sucks

cesium's! :(

and whose fault is that?

lmao

its like 14 fps

"Here is a more realistic scenario"

It's basically going to be used as a stress test

less battery

yes we can use in demo to say that straight paths is way better

oh yeah i saw some now

so there's little collision avoidance

jeez

The space isn't small enough

look

some do

wtf xD

do not curve

the 100 drones

wait

Actually I think that was 40 20

what values was that?

Oh boy

wtf, that weird sudden back up

Previous was 40, 20

This is what happens when search radius is 40 and drone radius is 15

them*

yh that's way too close and one of the stops

yeah

with reduced search radius?

headon they get too close

oo it's much nicer

I'm going to test it with head_on

Here's one with much smoother paths (reduced search_radius)

it seems fine for 100

It's still slow for 1000 drones

did we decide the test harness is still too slow with sending stuff off or was it jus tthe log writing that's the issue

Probably due to where waypoints are.

some crazy paths in there

here's a log

Oh I might have sent you the test file instead of the log

Here's a 200 second one

it worked on my end.

cannot read property length of undefined xD

its bugged for me

wait does it work

yep, I'll leave in running for 1000 seconds while I get dinner

oh alright, gonna move it up and see what happens?

that's only 100sec though

yep

or I guess you did since you have the file

sweet so it can write the log and download it all fine? so now just leave it to finish...

Paul for you're entertainment. Here's 100 drones confined in a small space

Yeah that stringbuilder fixed it

?

we can meet when Oliver is back and go work on it

Here it is again. Changed the structure a bit.

https://docs.google.com/presentation/d/10Yw4rmW5zx41Tl2DbfBBiQI8mBTnlxn6vOnRuQkGsb0/edit?usp=sharing

how exciting

meee :D

me after 1

to come in and properly work on the presentation

so who's free tomorrow

Cool. I suppose work on the presentation in the evening then.

i have an interview call at 16:30 so have to go after

Me

and then who knows

then there's a lecture at 2

whos going to 14:00 shit

then I assume we work on the presentation / code

I'm gonna head into uni again in a couple minutes

so whats the plan

https://www.overleaf.com/7599531csvgzqmjvdcw

Can you let me in pls? :)

paul and I are

Is anyone in labs?

Yeah

Just landed in ldn

Tmr at 10 right?

done

Or you message him

Unless he does decide to read this

That could be an issue

Hmm David like never reads the group chat and has an interview tomorrow at 10

David and Mery. Sign

Although I think section 5 onwards shouldn't be too bad

OK gl on that.

going home xD

ill be back at it tomorrow 8am

*clap clap

have had to do some serious editing, but wouldnt be anywhere without oli

alright im leaving it for the night boys page 20 and a half

My internet just dropped like a rock there

Cool

https://books.google.co.uk/books?id=MMeaI1TMvyMC&pg=PA7&lpg=PA7&dq=reciprocally+maximal&source=bl&ots=xb3SfSPCzp&sig=vN6QAo6AZEtjsmUD6-kVhpJV5z0&hl=en&sa=X&ved=0ahUKEwjiicrLxrPRAhWDI8AKHXs4BbEQ6AEIIzAB#v=onepage&q=reciprocally%20maximal&f=false

dw

right i found it

I actually did but then deleted it

Ok

can you also explain what "reciprocally maximal" is

I only picked the important ones

hahahah

...poor examiners lmao theyre gonna have to check all the references

being the first time he deals with this fucking bullshit

but by the time the reader reaches one line forward hes forgotten

you did, yes

I'm sure I did a couple of lines back

trying to make this accessible to someone with less than 200 iq

aight ill specify that

Capital Vb is a set of velocities

unless i misunderstood completely

no, VOb is a set of velocities

Vb IS a velocity

Vb is a set of velocities

selects a velocity from Vb

Also it would be +u instead of +1/2u

i have a question regarding section 4.4.2, what does this mean "given B selects from Vb"

so there's no way it could fail since it doesn't need to avoid it

is literally not on the drone's path

but that no fly zone

I get you

Similar to drones I mean

Manned aviation would be similar (Just another Kdtree)

No the test won't pass because the VO stuff only applies to drones ATM

I mean even so that test would always pass or are you saying you moved the no fly zone so it would pass?

Exactly

just... drones first

we can do that, obstacles are delimited as polygons

You have to default to our original algorithm unless we try to do a 3d version of the obstacle ORCAs.

how to crash into a wall 101

Nor manned aviation

No fly zones aren't I mean

Oh, that one isn't considered

I'm not convinced that no fly zone is in the middle of that drone's path

in theory if that's the only thing that we need to worry about being accessed and changed in parallel we can just lock it and then hand off new threads to requests as they come in from a thread pool

lol theres a lock keyword

http://stackoverflow.com/questions/11623039/how-to-make-objects-threadsafe-on-c

like if an object is accessed is it locked

i mean the c# docs lel

so idk if it's thread safe or not

I don't see any documentation

you can take a look at what methods you can call on the tree at https://github.com/codeandcats/KdTree/blob/master/KdTreeLib/IKdTree.cs

thats the one

correct

yes

this one? https://github.com/codeandcats/KdTree

seems to be a few they've done

what's the exact name of the package do you remember?

but id check concurrency docs just in case

well c# is supposed to be safe

but probably not

mm then it might already be safe with multiple threads

but*

provided by nuget package by still

it is indeed an object

if that exists in c#

or do a syncronized

if we did requests in parallel then we could just put a lock or something around accesses to it

I assume the kdtree is just an object though

/end

values are drone objects

keys to our kdtree are coordinates

child node side is chosen using the coordinate above it as a hyperplane

across MULTIPLE dimensions

each layer is one coordinate, and they alternate with depth, thus layers are xyzxyzxy etc

simple binary search tree

brief explanation of kdtree

as it holds all the drones.

kdtree

what common data structures are needed between 2 requests? kdtree I assume some list of drones?

so there are two parts.

Well, you can also serve multiple drones at once as requests come in.

so only 1 bit of it is actually parellisable?

ok

thats what i meant in the first bit by using a thread for each neighbour

yes def

oh orca lines?

They're done sequentially right now.

No I mean calculating the planes for the neighbours

KDtree takes care of that

neighbour calculations are already O(n)

Yeah so what I'm saying is that you do the calculations for the neighbours in parallel.

So it would be fairly quick for even 100 neoghbours

point of calculations is that linear programming is not computationally intensive and turns an NP hard combinatorial intersection problem into <n^2 discrete optimization

Obviously you have to do the linear programming bit together but you could calculate the planes for each of the neighbours in parallel

moderately sure

as in if you run this on an apache 2 cluster and set up auto load balancing it would split it pretty well

For doing calculations

but I thought something in the calculations was meant to be great for doing in parallel

actually apache does a good job of that itself

That would be just separate threads and a join command.

so we don't worry about that

handling requests in the real world would be handled separately by load balancers and all that shit

Oh you mean figuring out the planes simultaneously?

yeah that's what I thought I thought the idea was the calculations were parallelisable

one of the things is definitely multiple drones, yes, but i think even the vo calculations can be split

i believe they do it in two ways i cant remember exactly

section 4 of the clearpath paper

p-clearpath

nono hold on

Basically you can do multiple drones at oncw

So really the only thing we can do is multithreading for processing the requests

Well, our current system works on requests

what parts can we make it do in parallel or is it meant to do the entire thing in parallel I read the papers but I've forgotten what it said

believe me i appreciate the effort

yes i cant even imagine the nightmarre

Well, most of it

You didn't have to bloody write it

fucking neverending

done

does it not already do that?

signef

to avoid misdemeanors or assassinations tomorrow

you all sign cate now

i have a suggestion

ns*

I'll try to do that and see what happends

then the drone will update its velocity to be closer to the actual desired one

mb if the command is the same as desired velocity as seen by server, can make the command null

that could help for something, but I think not for this case

that could work

or mb we can make detection radius depend on the drone speed

just move them further apart i suppose

so things to change are constants and time horizon at the beginning of NewVelocity functino

i see

lel

the end positions are within the orca detection radius, so they are constantly trying to avoid collisions

i know

jeeeeeez

wtf is 4 drones doing

but for 4 - not quite

well, I thikn it works for 2 now

yeah, that's the app32, I think

Or changed it back so it works

You commited the version where they go kamikaze?

it committed my 44 test logs for drones approaching as well xD

ahahhhaha

ok, I'll commit this mess

doesn't work that well with 4 neither

kamikaze mode on

Now that's weird

wtf o_O?

then decide: "fuck it" and accelerate into each other

slow down

now drones approach

changed some values, behavioud shouldn't have changed significantly

ok, that's just strange

lel im stupid

oh yeah thats only in pythagorean formula

it doesn't matter

it's squared

lemme check

second coordinate goes first in subtractions afaik

are you sure that euclidean distance equation os right xDD

oli

um

LOL

basically I was creating new drone repository on every request xD

I guess I first decided to test something and then forgot that it was only for test purpose

I changed my fuck up to work properly

what exactly did you changE?

does it work for all tests?

that's a bit better

http://prntscr.com/dt3qyk

of course

lets just make damn sure it works first

As long as the state is accessed properly

If you can somehow parallelise the web server, it would work without much effort.

duly flabbergasted, in fact

13th

i will be impressed

if you get it done by the 14th

you can indeed xD

We're leaving it as an extension

coz it's super efficient if it's parallel which it isn't so kinda defeats the purpose a bit

or is that not something we're doing

if the translation works can we work on parallelising it?

everything up to 4.3 done

but it should be done easily by tonight

which is what ive ben doing for a long while

oli wrote it pretty fast so it needs improvement in some areas

bruce im only halfway through section 4

^

Having 4 eyes would be weird

On the report, obviously.

it doesn't hurt having an extra set of eyes

should I proofread it I assume you guys have already done that a couple times so it's probably a waste of time

Pretty much done. Paul's just tidying it up

how done is the report

I think so

was it 50

how big is manned safety distance

aight

yeah

is that ok?

the parentheses part

its a vector tho

wait

ok

yeah was just about to change that last one

and position is actually called location

velocity vector is just velocity

The state is not the same as what is logged

i can just put a paretheses like it has now

let me check

then why is there a thing under section 4.2.5 stating a drone's radius as a field

There isn't

oli what's the field name for safety radius in telemtry log

Misha have you checked if all the tests pass with the current translation of the stuff

they also use 1500 times the battery they have

Drones don't collide and they get to their endpoints.

insert large amount of quotes before and after works

But technically it still works

jesus though lmao good we abandoned that

well the picture is in the chat now so we all have it

No the 4 drones one was way worse

will was like "what the flying fuck is this

Keep that picture for the presentation

yes

yes lmao its hilarious

that's the drone_headon is that the hand written avoidance stuff

what up with that crazy path

aight

errored npm install works

ah alright

yep did that

you need to npm install first

sweet there's some already in there I see well got it the visualiser working now at least other stuff I'll wait til tomorrow

hold on

wait dont node

then type the name

have to put the json in the tests folder

how do I run a test, I give it the name of the test right but I assume for that I need the json first

cool cheers just give me 2 years while npm install installs all the shit needed to run it

Once you've got that you just run node server.js I think

You need node

given I'm on windows

to run the visualiser do I need to download anything or can I just run it from the folder somehow

The report bit I mean

Ok that's been added.

Also we'll have to include the apache license of the library (I think). Though that's the part that will be submitted Wednesday week.

hoooooly shit this implementation section is long

see what happens

Oh I would mention the actual library in the report (github/homepage link)

write a new test with 3

hm

no, its 4 going to opposite sides of a square

that its easy to miss something

also im not saying it makes it 2d im just saying

that's 2 sets of 2, no?

the one with 4 tho

oh ye true

10 are just sets of pairs

and why would that make it not three dimensional?

and 4!

there is one with 10 drones!

I mean there is no test like that I think

there is no such thing as three drones

like have you tried with 3 drones

are you sure the functions are three dimensional?

should I just commit?

and then people would get majorly pissed off

for now all the functions are in the drone controller xD

it would be plagiarism if we didnt say we are using their algorithm

no, we are using their library and saying so

its not plagiarism or something like that?

we are fine

are we fine with having code which looks exactly like the one in the library?

yeah so ideally I guess it would depend on the drone's velocity and acceleration capabilites I guess but we can just pick a value which is easier

that's right the lag factor

Not instantaneous change in velocity

If it was 20, then they would get too close due to acceleration

it's not like it's gonna actually be used for anything important

but we can just change it to make it look nice

so 20 + 10 was 30

which came to 10

20 is safety distance right and it was 20 + something to do with speed and acceleration

25 could work

So it had to be more than 20

I think when it was 20 there's a possibility that they intersect

im sure i asked this when i set it lmao

what was the reason for choosing 30 i forgot again

quite

again we can adjust it. We've got until Friday

I hope it can be done beter xD

Is it good as it is then? to me it looks like there could be less movement to avoid a collision but yeah maybe it's perspective

We adjust it every second as well

and that there is possibility of delay or command refusal

good point

and RVO doesn't account for the fact that velocity can't be changed instantaneously

can you push? so we can take a peek

cause I think they are bad now, that's why stuff happends

but the constants could be changed

I think I've translated everything that's needed

What impresses me about is that it's pretty much flat.

yeah. It could be perspective though

maybe if we change that then it'll decrease?

I think it might be about right give we set the distance to like 30 metres right

or else it looks further away than it actually is

They turn a bit far away.

let misha inform you better on that tho

and it hasnt been translated fully yet afaik

and its the basis for clearpath

rvo means reciprocal velocity obstacles

wtf is RVO and if we've translated it to c# now do I need the c+ version?

Maybe we can get a b4 and after for presentation

keep dem images xD

wait that looks good and everything lol

this one is the best so far, I guess

did you download a copy of C++ RVO

http://prntscr.com/dt2kkq

what files are where this shit is being done then I checked out to that branch but where am I looking and what am I doing what's going on

well technically I did my A-level project in C# and the js was webapps for me

0

Did I actually write kills? In all seriousness, how many us knew C# or js before this?

Clearpath

what branch are we working on

hilarious oli xDD

just found "broaden our programming kills"

hm

should be done by around 9

slow progress but leaving perfection behind

btw im on page 12 of report

gj mate

yeah, already asked oli, and deleted it

d

oh btw you can completely ignore clearpathcontroller file, not neede

well, when the drone starts avoiding collision I want to save its velocity at the beginning and use that as the desired velocity throughout the avoidance

cause i kept finding not enough info added to the agent

are you finding the drone model has enough?

and it doesn't detect collisions cause neighbourhodd stuff is not set up

yeah its a decent structure

and I use dronerepository again, cause I want to save some drone info

well, I just assumed what's already there is correct

whoops

Should have let you guys know that

still using drones as values then? might be pointless

i didnt even notice that lol

wut

also took me long time to realize that controller is created for each request, so I had to move kdtree to StateController

the radial saerch also detects the drone itself, so had add the ignore on that one

json log was again badly generated cause of some stuff

reference error*

also, I had some other sht

and that was that

and then my visual studio said there was a c++ reference and force closed

the thing with the returning drones

its fucking tricky i had the same happen to me

you sure you got the math right?

but that's not for orca says should happen :)

That sounds kind of cool

other drone just goes straight

one drone goes 90 degrees of its course, after some time continues to destination

you mean bad visually?

nothing too weird

well, bad, but fine

this one looks fine

ok, that's the latest log

Oh you have to delete the old log file I think

most of the time

so I run test harness, after first test I kill it. The log is generated, but its bad

Check C:\\Logs

It should have downloaded a log for each test

so to get a proper log, I have to wait until it finished all the tests?

In the middle of it

I think it happens if you restart the test harness.

http://prntscr.com/dt1qhm

Why does it sometimes not write to log properly?

well i WAS working on that but the report is kinda coming up xD

I tried to fix that, now they pass through each other (like no deviations at all), then one of them suddenly turns back and sprint to the other one, after which they go together

not my fault nobody is actually sharing what's going on I didn't even know misha was working on it

Do they pass through each other or do the red spheres just get close?

yh but I thought you were doing that or we were just gonna use the c++ and find a way to do that

the c++ 3d copy of rvo collision avoidance to c#

bruce you are completely out of it arent you darling

What are you translating?

ahahahha, so I tried to test the thing after my translations. Sometimes it fails, sometimes it passes even though drones pass through each other. This time drones approached each other, then one of them turned back, so that they both started flying in one direction. After some time they timed out

YES

I'm just gonna assume it's 10 then

Ok here's an outline for the presentation https://docs.google.com/presentation/d/10Yw4rmW5zx41Tl2DbfBBiQI8mBTnlxn6vOnRuQkGsb0/edit?usp=sharing

David has an interview then so I guess if we're still there later on he can join us otherwise we'll have to proceed without him

sooooo are we meeting at 10?

I think they were trying to make labs seem brighter

the photo on this page is so out of focus its actually sad

http://www.imperial.ac.uk/computing/csg/facilities/lab/

But yh When are we meeting

It was earlier and clowed later I think

I think they moved the times last or this year

Wow Disappointing

lol

its 8

I thought it was 6

when do labs open?

no 5am better

Sure so like 6? In the morning That cool

I'm fine with anything

So all but David have confirmed Good enough 9 is too early I think Shall we say 10, 10:30?

Aw they took my frigging liver

You bring it up

how do you drown a shark in water

tomorrow morning should be fine

fuck you

...once

hey, i went and talked to them :(

yh but also people who use the careers service are meant to have better cvs and get better internships but who the fuck uses them

Apparently groups who go do it end up doing much better than those who don't

Cool. Don't forget that there's a rehearsal thing on Wednesday

so we need to get hold of Misha and David and see if they can do tomorrow

yh still we should meet tomorrow to start the presentation and get everyone up to speed

room 145

10:00

oli's right

ye 13th Jan

Presentation is on Friday

I think one meeting isn't enough for tgat

thought it was the 14th

saturday (?)

9 is a bit early plus the report is mostly done right, I was saying meet to do the presentation considering that's due, what, Thursday?

9 or 10 would be fine for me.

so you guys decide

im happy to meet at 9 but available for any time after that

Im back tonight, tmr is good for me

meet early, report is due at 12 if im not mistaken

not necessarily in that order

and also decide a time

yeah, just need to get in touch with everyone in the group

That's at 2 so we've got plenty of time beforehand

something \about presentations

yep that's definitely England, cool I'm gonna assume Oliver is David is Should we meet tomorrow? there's a lecture anyway

im in uni finishing up the visualiser latex

Are we all back in England now because I think we should meet up as a group and talk about what's going on / do the presentation

The only thing left is the visualiser part.

Ok VO stuff is done for the report.

Ok I'm done for tonight. The only bit left are a couple of screenshots and the VO stuff.

precisely what im here for! :D

We're 30 pages boys! (We'll definitely need to cut some of this down.)

yes I saw, not bad

well guys, we're going to reach 30 pages easily

I'm in the airport now

i know that

no internet ✋

So stuff like one of the task boards and the build server stuff.

Could someone get a couple of screenshots from VSTS? (I'm still logged out)

Yeah, I'm just doing the images now and finishing up the other algorithms stuff tomorrow

Makes sense

etc

make sure it doesnt look like were talking about the same stuff again in some others

and ill def read entire thing and rephrase/extend some parts

So you're just doing the visualiser part, right?

just curious im just gonna do report tonight and tomorrow on the plane

mikha how are you handling the timestep in computenewvelocity when translating? cant figure it out...

Just that and images (which I'll do tonight).

Ok the only section left is the ORCAs one.

that way we can all edit it

If you're going to do that you should make it a google slides

agreed

yeah but I mean we can plan the presentation and make the slides and just leave placeholders or a slide that says "demo"

but priority now is report

so we better get a code skeleton working real fast xD

point is we have to give a working demo for presentation

We've got the entire week so I suppose then

Well when are are we gonna work on it Because that we should do together In labs ideally

You mean has anyone started it?

What are we doing about the presentation?

Should be able to get it done tomorrow.

Ok three parts done on the implementation. Three parts to go as well as images.

I would cut out the json bit

Good on the exec summary so far but I wouldn't go even that technical for the test harness.

OK back in London. Should have most of the rest of the writing done tomorrow. The rest of the writing and images should be done on Friday.

That's fair enough too.

ok, I think my point is not valid, since this would still probably be better than other algorithms since we wouldn't know the actual desired velocity for them neither

So if there turns out to be upcoming collisions, the algorithm will give a velocity as close to possible as its current one.

Assuming no upcoming collisions, the drone is trying to get to its waypoint, Right?

it*

by fuck up I mean if might steer quite far from where it should preferrably be

ok, wait

Ok let me explain

they kind of achieve the goal

well, I mean in their example with people or some other things

If it's continuously avoiding collisions it would never get to it's location anyway

so if there are many drones, so that the drone has to continuosly avoid collisions, this will fuck up?

because always tries to turn towards the waypoint anyway

We're treating the current velocity as its desired velocity

so that part of algorithm wouldn't work?

So it can't know the desired velocity after the robot is deviated from course

it doesn't

does the server know the waypoints of the drones? Cause it doesn't right?

sorry if that was discussed before

btw

as in, why the fuck its before my individual project ends

anyone know why postgraduate graduation day is in may 2017?

because they're annoying

Why is it such a short period of time? Idiots

It's from Monday to tuesdau

it doesn't seem to work for me so idk if it's actually gone live yet

we're gonna need a reminder for the webPA thing

I think I can manage the algorithms if you manage the visualiser as I have no clue when it comes to the visualiser.

offline edit

ill take care of writing it over there

leave everything on new algorithms half done

dw we will do a good job on that

Unfortunately we only have up until midday on Monday.

Ok I should have the report done by saturday.

xD

lol

saturday 23:59

do you mean Saturday at 0:00 or Sunday at 0:00?

dw i got it

its on visualiser branch, just checked

short psa and my plan for the next few days: keep translating stuff tomorrow the 4th, then make suitcase and off to neighbouring island on the 5th to visit family for 6th national holiday (with NO internet so will take physical copy of everything including everything in the report until tomorrow night) then arriving london 7th midnight

If not you can go to a previous commit and download as zip

is it there locally?

might be a merge gone wrong somewhere along the line

ill fix by the time i go to bed. weird

There's only node_modules and tests

Nearly all of the files in the Visualiser directory are gone.

I found an issue with the current branch and I'm not too sure if it's intentional.

Another section finished (Evaluation)

Another section (Conclusion) done although it requires a few images.

*right

Write, section on extensions done.

https://www.overleaf.com/7599531csvgzqmjvdcw

Ok could someone (other than Paul) please write a one page executive summary.

xD

unfortunately this is real life

win for everyone

then they'll just give us an A* for finding them and not give a shit about us anymore

would be hilarious

writing all these papers on it

maybe we should just tell Altitude Angel to go work with these people

but it has pretty much everything in it

ayyy

fuckin computeNewVelocity has 341 lines

It's definitely the least useful of the papers.

I only referenced it because it has extra information on something or other.

The underlying maths with VO are similar

good question regardless

also is there a point in reading the ORCA-DD paper it's about differential drive but drones are Holonomic

Velocity Obstacles are the commonality

They are different. ORCA is like a more advanced version of clearPath that prevent oscillation.

it even has the same pictures

it seems like the same thing

is there a difference between ORCA stuff and clearpath?

We've got the old version anyway

cool

but im changing a bunch of stuff tho

nono just rewrote to use normal one

actually you have to change the call of putBytes in Drone.cs

You'll have to edit it in Communication.cs in test harness if you want to still use that controller.

Also if you're still doing it in ClearPath controller it won't work because no messages are sent to that controller.

yes

Trying to adapt the old controller to the new one

what's going on here

I'm confused I thought we had the detection stuff done

It would've thrown an error on the first PUT

if it didn't work then we wouldn't have gotten anywhere

you sure that works then?

yes

this thing

?

Drone drone = new Drone(id, cartesianLocation, data.velocity.ToObject<Vector>(), data.acceleration.ToObject<Vector>());

46 (I think) in DroneController

It's one fucking line

im just gonna make a simplified drone class to use with tree, fuckit

okay

Trust me it won't work when you actually run it

i mean no errors

it does work though

data includes the time too.

Check DroneController

That's why I didn't do it

data.ToObject<Drone> won't work

data.ToObject<Drone>() *

or at least data.ToObject(Drone)

cause if not none of this works lel

im hoping

the "data" thing in your put is the actual drone

btw

alright small progress update, we have a working 3 dimensional binary search tree mapping drones' positions to their drone object, getting nearest neighbours within a 30 metre radius and using their velocities to generate first parts of VOs

that's what I was thinking

exactly

position, location, id, velocity but that doesnt matter

you can just compare uids.

what info does it need?

how do i test this shit

just realized we might need to override equals() and hashCode() on drones so the kdtree can find their info

We can try with 30 and increase if we need to

xD

what's the final verdict on that

which was 40

I was just copying what we did before

of course

yeah but we don't wanna be too liberal otherwise it'll break down when there's tonnes of drones

40 works too, they'll just be further apart.

So yeah maybe around 30

so what, 30?

but ok so that plus the safety distance

yh but the point is that if it's outside that range then it won't collide in that timestep so you can ignore it which is detection

This is just for detection, not the main ORCA stuff.

No because that would mean that the spheres intersect which we don't want.

which would be like 10

but the max velocity of A, plus max velocity of B, multiplied by timestep

isn't the distance to search this: (vmaxA +vmaxB)τ copied from the paper so didn't quite go

yep

Well you'll have to do the visualiser bit

gj tho

ill definitely go over the whole thing with you on 8th/9th

Two sections down. Four to go

Ok that Introduction section is done. Might a read over and a couple of images.

Love to see how it turns out

cool

easier to run nnl searches on it

just changed my mind, gonna make KDTree keys into double arrays and values into my own class

That we have same time trying to solve an LP with 100 constraints

Land would be dumb because it could still collide

yes

Actually hover

I suppose return land command.

i can make it so that if there is more than 100 neighbours something happens if you want

aight

40 metre sphere

So we can preemptively prevent collisions.

Because they should not come within 20m

Actually 40m

20m

Twice the safety distance?

btw how many neighbours do we want to check / what radius do we want to search

due to my irresponsible nature xDD

I was gonna push code today around an hour ago and its gonna have to wait 😂

like its nice cause i rarely go out in london but it gets to the point where i feel bad cause i havent done shit on this xD

fuuuck dude ive been playing social life way too much this holiday

Most of the meat is in sections 4 and 5

Will start looking at the Introduction section as well tonight

Requires screenshots but it's just shy of two pages.

https://www.overleaf.com/7599531csvgzqmjvdcw

Ok the planning part write up is done.

Our plans probably changed so much anyway

I managed to find a copy of the justification for Scrum (lol) but not the plan

Could someone work on the executive summary? Basically a page on what the project is and why people would use it.

got it

so for 47 to 56 would be your kdtree stuff

old copy is safe

the old copy is safe now right?

so can i just randomly replace stuff in dronecontroller.cs? so far ive been writing everything either outside the stuff git tracks or within another controller

um no i cant get a copy lol

because that request is for the one drone sending data

right

yes

which i think is the intention

and also as far s i can see thats a single drone being added

ill get a copy for you in a sec yes

So you would keep line 46 and replace the other lines with your kdtree stuff.

lines 46 to 56 in drone controller

You should be able to get it on cate

Oh btw Paul could you get a copy of our planning bit from the first week or so of term.

I'm not too sure it would work though

Why a different function?

do i just make another function like [HttpPut] public HttpResponseMessage Put(int id, [FromBody]dynamic data) {

what i dont get is when that happens

like i understand the copy in the repository is being changed

in your drone controller, ive been reading through it and i dont get where state is updated

Goddammit

xbox

?

Which controller

😂😂😂😂😂😂

cheers Paul

Orca Orca with bigger titties

iximperial.visualstudio.com

it's like iximperial or something right

yo what's the link to our repository thing

just gonna have lunch

im not getting how im supposed to listen for drone pings in the controller so if you can clairfy that for me

I think I read the clearpath one

start with those

most important ones are unc, clearpath and orca

they're all related as well

yes

Are they all about the collision avoidance stuff?

Again, doesn't help with the cone stuff

oh i didnt read the airplane one

good guy oli

Here are the papers btw

absolutely xD

Yes. It would also be useful for the report too.

As in would that be of any help if I manage to understand the paper and how all works?

So do you think we will have problems and lots of debugging with that thing?

Right

righto

It's done there in DroneController

and which file receives them?

Excluding the type

You know the logs? Whatever is there

That's the only time the state of drones are updated

the requests are little jsons with what exactly?

Yeah

btw just to be clear you want the kdtree on server updated only whenever a request comes in right?

okay

It only needs to be committed to run the automated testing.

It doesn't matter, fortunately

but it doesnt really matter

in fact i dont have a clue how to make it generate a dll in the right place tbh

changing the dll ref seemed to work xD idk if i pushed the change correctly cause of the gitignore

Haha, now Paul is having same errors I did

I'll start writing up the report tomorrow. Will be available for questions.

cool

cya boys

will keep going tomorrow

alright pushed that change to utils so i can use toDoubleArray

thats ok i guess

id have to force the commit with -f

ok

or do i have to move it every time i edit utils

Yes

Just remember when you commit, add bin/debug/utils.dll

is it going to work from now on

i have no idea what you mean mate

I only moved it to Utils/utils.dll because I was too lazy to force commit it.

I don't mean change it back

Leave it

is there no way to automate that?

but

i got it

confused

im so

I think it's that

Try then on tool bar Project -> Add Reference

well it doesnt let me edit the path

if you build debug

VS does that for everything

That's where they build to by default

and why on earth does it build there

Meaning if you build utils

No because utils builds to bin/debug/utils.dll

(?)

surely thats correct

it references Utils/Utils.dll

There is then in its properties a path to utils.dll

Select the reference under the project tree.

how does one do that

What you can do is change the reference to the one in bin/debug/

Not the one in bin/debug/

At the moment, the reference to utils is referencing one in a different folder.

is that just how c# works?

why is the change not propagated

hold on wtf

Whatever you do utils.dll has to be committed.

And remember to commit bin/debug/utils.dll

Or change reference to one in bin/debug/

Go back a couple of folders and there's a utils.dll

move dll into what?

You have to rebuild utils and move dll into parent folder

?

You mean server doesn't recognise it

its not recognizing anything new i type into vector.cs

getting the weirdest fucking errors tonight lel

It comes from newtonsoft

Don't know dude.

like how do you properly include newtonsoft

when i rebuilt it gave me an error lel

what package does jsonconvert come from

create a file for the fuckton of constants

What are you trying to do?

you err access it.

btw how the fuck do you access a const variable

but its not the main focus so if you want we can forget it for now

idk its good practice i think

they did say to log everything for auditing tho

I think having the server say "Drone cleared of collisions implies it"

It's kind of too much information at this point. And the server cannot tell whether it's been changed.

Can't we just observe it

Why though?

what i mean is, if the drone changes direction, whether that be the test harness playing with it or it being sent a proper command, it should be logged

Is that what you mean?

So a message like "Drone (id) cleared"

If it was sent and this time one wasn't it sent you could mark it as "Drone cleared"

What you could do is keep track of whether a command was sent the last cycle.

But the server doesn't know any of that.

welp maybe that should change, just so we can say that gets logged as an automatic redirect by the server

No because we're assuming it's some guy pointing the drone to the new location.

as in, is there a ping somewhere indicating the automatic redirect occurs

is that action logged?

That's why Will was concerned that our drones weren't given enough commands.

Otherwise none of our tests would work.

Yes, if it needs to.

yeah?

and it goes ok

so it generates a nw vel all by its little self

That's what it does right now

So it goes "Oh, guess I'll turn directly to my next waypoint"

Meaning if there are no issues you don't tell it anything

Already in the test harness

because the point of clearpath is that it recomputes a collision free velocity from an existing one

Drone does it by itself

but somehow we need to redirect the drone to next waypoint

alright we can do that

waypoints I mean

Because we were assuming that the server doesn't know where every drone wants to go.

I thought we were going to assume that the current velocity is the desired one.

right so a drone arrives and i need to get it into kdtree, just wrote that

i just need next waypoint not even last

inf act

cause i need to calculate desired velocity

it would be easier if i can pull first and last waypoints

right even so

So: drone sends telemetry -> updates kdtree -> generate command

In terms of initial positions, it's only when a request is sent.

It's initialised if it's a static variable (which it will be)

does the server get sent all initial positions at some point?

alright so how do i initialise it

so plain aggressive?

Like active aggressive instead of passive aggressive?

kdsunlight im dying

xDDD

more or less what oli said, but more aggressive

bruce

Bruce, your shitposting, while humorous, is not very productive.

And make sure it gets lots of kdsunlight

And you water it

Like we do currently with neighbourhoods

In the kdgarden

Every time a request comes in from a drone, update kdtree

Keep a kdtree on the server, surely?

cause in that case i dont really understand where we keep a kdtree

So we have no edge

Well it doesn't actually work atm right We haven't translated it or whatever

is it gonna stay there?

oli, you said your test parser was somewhere in test harness

correct

Doesn't matter now. We have the advantage in the drones vs drones department

But we could just use the code we have already for that.

There's that too.

yes, and not fairly easy to construct VO for the plane

and we re going to completely ignore that for now! :D

I mean, if you know the dimensions for ellipsoids it would be fairly easy to calculate the distance.

Oh true dat Except they did say ideally planes for example would actually have a conical safety area

which would be the real-world scenario seeing as safety distance for planes is massively increased in their forward direction im guessing

so ellipsoids of any other shape arent considered

fortunately we were told to consider spherical safety distances

We just have a constant safety distance.

radius is not the body of the plane

Or is that not an issue

I guess one thing we may have to address in our report is that it works with spherical agents but planes and shit aren't spherical

so it's only sent once for each drone.

What we could do is have a post statement that gives the radius, maximum velocity etc with an id.

righto

I think I noted it in the outline already

yes, exactly

Radius is constant ATM. We could note that as an extension.

okay fair enough

and we can keep radius constant but say its modifiable

It's changed when the state is updated

just wondering if i can get away with rewriting agents with just position velocity and radius

no, not at all

Why is it causing an issue?

Acceleration vector is used for genetic algorithms. It's never changed, though.

either*

wither vs is stuck or i cant find references xD

oli is the acceleration vector actually changed anywhere?

Can't believe that's a genuine song

https://www.youtube.com/watch?v=4Jvn36OVdKs

thats fine ill check in a second

you might need to redownload it

ah okay

if that's what you mean

it should be kept as it's in visual studio project settings

idk how it works on git lel

so can we use the same nuget package?

pull now

noice

already done

ill do it if you want no problem

two seconds

Not yet

so i can pull it

is there a new branch now?

definitely

We should also keep the stuff for the noflyzones BTW

So a request comes in -> new velocity calculated -> new velocity returned as a command.

Where a time step is 1 second

It will be done by the requests coming in.

That's fine

well... allegedly we have to recompute velocity at each time step

Do we have to count time though? Other than tau from a particular period in time.

i believe

as long as there is something somewhere counting time its all ok

Rather not rewrite the test harness as well.

including running a copy on test harness instead

we can do it however we want

i agree lel the only reason it exists in their code is because its waaaaay more generic

Because we're using "actual" drones.

The simulator sounds like a really dumb idea.

need two singletons, one is rvosimulator and another is kdtree

We should only calculate the new velocity for every request that comes in.

ah oli this is where the simulator comes in

they LIED

no it did not

nice "fuck off" gif

there we go

What happened to them running each team's tests against the other's test harness Did that ever happen

http://gph.is/1Th3Spl

If we recalculate all the velocities every step.

So basically it looks like we have to simulate the drones on the server too.

fuck chatley

side note does anybody have a clue whether to use the DLL or the static version of c++ code for reference

We have a bunch of tests anyway

That isnt tdd Write the test, watch it fail Make it work Did Chatley teach you nothing??

if it happens to be the same, well... idk xD

"We will recompute this velocity for every agent, every time step."

lemme just link a sentence from paper

Is that the same case now.

At the moment drones only get commands on a request.

aight but how about writing the actual thing first

can you rephrase that question oli xD

Well write a test with 100 Let's do it

so... heheheh

and realistically our demo will be 10

M8 it's just pintos How hard can it be

i promise its fast enoughfor like a 100

i mean were on a clock here xD

we say it CAN be done

we dont

But do we do it in parallel tho

like to the level where you can run 25000 drones at 100 fps on a server cpu

Wait, so we're not just giving it on every request?

and while that may seem slow one of the main points of report should be that the thing is highly parallelizable

we loop through every single agent in simultion, check their neighbours, construct VOs for all of them, move them, and get new velocity from the boundary of their union

We have it. It's just converting it

Then it's just the avoidance stuff

That's another cool thing to mention.

Oh sweet

kdtree takes care of it

They can be done separately imo

detection is done

you only need to read first 4 pages

Nah detection is which ones should we check Avoidance takes that list and calculates what should happen

oli was faster cause internet in the canary islands hehe

this 'un

'old on

detection and recommendation are kind of together.

Which one m8 There have been many

have you read the paper

bruce my friend

What's going on

Who's working on what

inb4 http://stackoverflow.com/questions/1519006/how-do-you-create-a-remote-git-branch

I mean Do we know what's happening Like So we need to do avoidance and detection Are we splitting that

name it clearpath or something

aight u got the console open

A new branch

so are we just working on master?

Pushed

Ok created a new branch to hold the old server.

id is int right now

int? xD

but the id i have no clue

Oh ok

so i think it fits with double

the key is a coordinate

for now i have double, double

id and drone?

we need to think of what the keys and values are

there is a c# nuget package called kdtree which is good

and to add a dimension the only thing you do is literally make the levels go xyzxyz instead of xyxy

As updating the state and GAs are in there

Oh btw. We shouldn't ditch the entire server. As up

I believe Which is definitely useful

basically a glorified bst

the only thing it does is make nearest neighbour searches massively efficient

Or the n nearest neighbours

It's a way kf finding the nearest neighbours withon a certain distance

kd tree is actually much simpler concept than i thought

We just check for distance?

So basically the kd tree supports the collision detection side?

so thats fine

cause we need one for position vectors

right

It's whatever origin of the utm zone is

btw is there an origin for cartesian coordinates somewhere?

Just do it on a separate branch called server_old

xDD

great minds think alike m8

we need to make sure we include an image of how the current paths look

SO we can say "here's what we did before"

yes exactly

Also for the report.

considering the only thing they use is the vo thing

Yeah keep a copy so if it doesn't go well

xD

if you guys agree

Great...

id say so

So neighbourhoods are kill?

There's still important stuff in there for interfacing and that.

but still xD

like keep a copy

i agree with bruce, server is going down the drain as far as im concerned

We're rewriting but simplifying it

ah also their vector is quite a bit simpler than c#'s vector, so we might as well use it xD

Yh but the collision detection and avoidance is the core of the server The rest is just handling the receiving and sending

We don't have that much. Unless you're talking about the test harness one

drone.cs

nono ours

You mean the agents in their code?

and i think we dont really need all

So is the new repo just to separate it from the main codebase before we put it in?

cause the current agents have a fuckton of stuff

we might need to simplify stuff

The "generateCommand" bit

it is better as far as we can see in more or less everything

Exactly^

Algo*

Just the also

I think not completely

I.e we're basicallu rewriting the server

as to the question of whether we start tomorrow, ill be pushing some code in the evening if thats useful, a new repo with some predefined models based on what they're doing

It's theoretically optimal

So we're rewriting the collision detection and avoidance algrorithms?

Is it better in all aspects?

It would be more if someone has nothing to do

also, unit testing is imo not so important atm, should focus on getting the algorithm done

some comments - bruce, to expand on what oli said, we're rewriting the collision detection algorithm basically from scratch to use some research from UNC which theoretically gives even better results than what the other group achieved.

Images are key

Also, the 30 odd pages won't all be writing

So other people can work on the report.

On the collision thing, maybe 2

How many people can actually work on that simultaneously?

We might want to write unit tests for the test harness as well.

Yeah

The visualiser is finished then completely? We're just perfecting the part where we find new velocities for possible collisions?

We're trying to smooth it out.

If you saw some of the visualiser things, especially for 4 or 5 drones, the paths are really weird.

To get you up to speed. Basically we'll be trying to give better recommendations with a different algorithm.

Which is great

I know

Was my point

I'll have a computer tho

Cool. But I won't be back until Wednesday

In the slightest

But I also have no idea what's going on

Well I'm back in London tomorrow

So are we going to try implementing that library tomorrow?

but still gj mikhail xD

man, two spaces is weird

Added subsections to report for more guidance

Already*

Lol, I completely missed sun. I went to bed it was still dark. I woke up - it's still dark

Obviously there's nothing in it

https://www.overleaf.com/7599531csvgzqmjvdcw

Here's a link to the report on Overleaf

Impressive

may we not fail on our quest for grades

officially ded

got home at 10 this morning

Happy new year everybody

Sorry guys I was on holiday w my family and had no internet at all! Happy new year

extremely late happy new year from the canary islands

Late happy new year from Russia

Happy new year from England

useful resource

Also goes a little bit into linear programming

like i said, there's nothing on the cone for 3d

https://software.intel.com/sites/default/files/m/d/4/1/d/8/UNC_collision_avoidance.pdf

Now this one is good. It gives pseudocode of 2d library

formatting is more consistent

less effort

its hilarious tho id have used a txt

no ofc not just kidding

I'll convert to pdf if you need me to

xDDD

in fucking docx

Here's a quick pseudocode sort of explanation for computeNewVelocity.

Interesting btw is that the alpha mentioned in one of the papers is the 1/2 that u is multiplied by

Well I can't explain what every line does. But I know that the linear program part for example definitely works.

if you truly understand how theyre doing the full process i would appreciate an explanation. ive been looking through their stuff for ages and only get about half, i dont know why anything works

the other stuff is explained by those papers

Yeah. Fortunately it's still only a tiny part of it.

but i doubt he'll answer in time being christmas

ill send tonight before dinner just in case some more pop up

Oh another question would be why timestep is used for a collision but timehorizon is used for a non-collision

clever, actually

yeah so his lastname is snape hes just putting that to avoid spam im guessing xD

He also authored the paper in the 3d part

lastname@cs.unc.edu

Holy fuck that's funny

k

snape@cs.unc.edu

wtf is his email

any other questions?

Just where that quadratic comes from.

aight ill drop him an email

Like I said, it's the 3d version of projecting onto the legs of a 2d cone

The only thing that isn't explained is that cone bit anyway

I mean he would be the best bet seeing as he is the owner of the repo

HAHAHAHAHAHAHAHAHAHAHA

Wasn't he killed by nagini?

man i cant get past the fact his name is fucking snape xD

we could try snape

no response

emailed stephen guy

about a week ago

i did

*hurt

It wouldn't help to ask the authors

yeah... me neither

Other than it's the 3d version of the 2d cone.

still haven't found any information on how they do the 3d bit, unfortunately

Another ORCA paper http://gamma.cs.unc.edu/ORCA-DD/ORCA-DD.pdf

This one ^ explains more why they project on to the legs. The cone is just a 3d extension of that.

The paper on ORCA http://gamma.cs.unc.edu/ORCA/publications/ORCA.pdf

Ok guys, we are using 2 space indents from now on, I will change everything to match that (for now we have 4 space, 2 space, 0 space, 2-2 space indents)

Ok so w is the vector from the relativeVelocity to the closest boundary of the VO.

ok maybe not everything

it's the paper that the 3d library is based on

http://gamma.cs.unc.edu/S-AIRPLANE/S-AIRPLANE.pdf

hey this should help explain everything

w is always relativeVelocity minus some multiple pf the relative position

timehorizon is minimum time

It's just figuring out how they get it

For a cone it calculates a specific inverse time

meaning that the direction of the plane is also the same direction that the plane is away from the velocity

Just so you know, u is just a multiple of w

xDDD

ok. mb not

this is funny

ouble units = a * a + b * b + c * c + d * d + e * e + f * f - 2 * (a * d + b * e + c * f) - 1 * 1; double ts = 2 * (a * x1 - a * x2 - d * x1 + d * x2 + b * y1 - b * y2 - e * y1 + e * y2 + c * z1 - c * z2 - f * z1 + f * z2); double tSqs = x1 * x1 + x2 * x2 + y1 * y1 + y2 * y2 + z1 * z1 + z2 * z2 - 2 * (x1 * x2 + y1 * y2 + z1 * z2);

I am looking at our code for now

ill probably just write down it all in one go

no problem

gotta go back at 5, but just write stuff down here and i can try and answer

yep

line 136

So the obvious first bit is that the spheres are not intersecting on the first if statement

yes im sure of that one

The radius is like our safety distance, right?

welcome to the club xD

It's just a certain part of it. I'm just trying to figure that out

I don't mean that the plane is the whole VO

and if you look at their definition it makes sense

it says that in the paper im not crazy

no, the VO is a cone in both 2d and 3d cases though

That's why it's a plane instead of a cone

From what I understand

the plane is a velocity object for a certain period of time

yes

the basic premise is tha it's the set of all potenially colliding velocities

(right?)

except the thing that represents the agent collision is a VO

Not that the plane itself is the agent

Sorry I mean that each plane represents how the agent could collide with each of the other agents

how is that plane defined

First few or lecture notes should help

why? I thought each agent was a point with velocity and radius

Each plane corresponds to one of the other agents

OK so onto understanding it. You're trying to solve a linear program with the planes created as constraints

a slim chance indeed, considering I hadnt even heard of linear constraints until yesterday, which was when I realized thats what theyre doing

Well there is a chance that while you are translating you will get it

xD*

yxD

bad strategy mates

Theeen you try to understand

how the fuck do you fix

but if somehing doesnt work, which it wont

Right

i mean idk, yes, you can literally translate the maths over

ok then

trust me on that xD

I dont know about you but I find something I dont understand extremely hard to bug fix and code in the first place

I get that we have to understand it but shouldn't we try implementing it first?

(and in report 😊)

good idea

yeah that way we can get destroyed in presentation

That's the point. We don't need to understand it fully if we're just converting it to C#

why does it work

alright, so explain the whole process to me from start to finish

it's just vector calculations

I get that, but it's not like C++ is using esoteric shit

so we can implement it in c#

we need to get what the c++ copy is doing

the c# copy is 2d

man

Like I said, I don't see why you need to modify any of this.

It's just finding the time period

*matter

The a b c shit doesn't matte

okay, even so, I still have no clue what its doing

I think the reason why it says projecting on a cone is because it is a slice off that cone

t looks like an amount of time to me

calculateNewVelocity

linearProgram3?

which function are you looking at?

https://github.com/snape/RVO2-3D/blob/master/src/Agent.cpp

what file sorry?

also c = 2c and a = 2a in the actual quadratic equation

*equation

i'm not sure if this helps but t on the project bit is a solution to a quadratic requation

compare the 2d and 3d versions

but the functions are different and the linear programs are different

exactly

a 2d one

no the legs also define a cone

that cone is a 3d version of the "legs" bit

that's it

Literally replace the project onto legs with project onto cone

the 2d copy of C# to support 3d

what would you need to modify, though?

also I need to know why the computed velocity is not, as they say, the closest to the boundary

so yesxD

well, if I cant understad it I cant modify it

ok it's an understanding thing

3d cone I mean

if you can explain to me why the constraints they give magically define a cone and how they calculate intersections between them I'll be happy

looks like standard vector stuff

What part of the cone bit is difficult?

and yes I agree its weird but if we can figure it out its just a question of writing a small parser for it

also there is an obstacle object somewhere

did you see their c++ stuff?

yep... xD and the hard one to implement

I'm guessing that's the hard part to understand?

Also 3d is basically the same as 2d except you're projecting on the cone instead of the legs.

basically, but the code they use is weird. Probably because they represent them differently.

Wouldn't that just be the zone itself instead of a cone?

Because I have no clue how they're doing static obstacles.

I think when adapting this we should only do other drones first and then add noflyzones later.

That we know of

Nah they're pretty chill. No Orca has killed a human in the wild.

Well The world is 2/3 water right So they own most of the world already Once they leave We're screwed

yet

That wouldn't work. They can't even leave the water

Planet of the Orcas In cinemas 2017

Well 2d But looks 3d

It's a 3d orca

constructive

The number of lines it fails on is always less than the number of orcalines, so they have to them solve it three dimensionally.

If it is unsuccessful it returns the number of lines it fails on.

If it's successful than it returns the number of orcalines

So they try it with 2 dimensions first.

Solving linear programming 3 dimensionally for 2 dimensional vectors is ALWAYS successful.

Ok I can explain this.

brb

check the code for agent. computenewvelocity i dont thibk its that simple

It does that by solving a linear program

*velocity

Basically linear program puts in result the new vector

they have 4 total

but it has to be to fit with the format of linear constraints

their way of defining cones is also quite annoying

question is why xD

phi basically projects the direction (v-p) along another vector mu

it helps to draw it out

but we can point it out and explain how

found phi btw

which we arent gonna do

yes, however the point of the whole thing is that the main loop can be very easily parallelized

The way we will have to deal with it is only considering one drone at a time anyway

i say we forget the intricacies and deal with ideal drones, then fall back on what we tried for other things

Mb it can be simplified since we have lots of uncertainty anyway?

And also velocity can't be changed immediately

this is something i brought up when proposin it

So if they don't Could cause issues

It*

correct

This is true If assumes all agents follow their trajectory

page 3

so reflection is the exact same

because when they explain minkowski difference they use ref point in middle of agent

Would drones which ignore commands or the fact that data arrives at different times or other things not cause problems?

If so which page

is that in the paper?

problem is idk where they consider origin

subtracting a position vector from a velocity vector then dotting with something

and the secondary, slighy but not very much lesser problem is the cone itself. they use a Phi operation which appears to be something along the lines of using a dot product for some obscure reason

so you want to understand how the solve linear program bit works?

even after spending hours drawing examples xD

oh ok

which is done using linear programming and I have absolutely zero clue how it works

the very big problem is the FVO intersection

that's what it looks like.

basic procedure is 1. store everything in kd tree and loop 2. get nearest neighbours of agent 3. compute FVO to each meighbour using a mixture of minkowski arithmetic and linear programming 4. recompute velocity to point to closest point of intersection of FVOs boundary and do that while there is at least 1 agent which has not reached destination

k

and in 3d it would be the planes defining the cone I think

optimal reciprocal collision avoidance lines

The lines of the cone?

what does orca lines mean?

ive already gone through most of it

if anyone doesnt get something, ask it here

hahahaha lmao at the stabbed comment xDD

and we will also have to do modifications for it I guess

right i'm going to attempt to understand this code

Besides I'm from Basildon, what you get from your neighbours is stabbed hahaha

Well Generally people don't have spare computers mr "rich fancy spare computer man" :P But meh I'll be back on the 2nd

that is some bad luck right there xD

man how tf do none of your neighbours have a spare pc

quite

I'd help but looking at code on your phone isn't useful :') Even if we don't get it working we can write about it in the report

btw bruce! you may be right about writing a wrapper for c++. no idea how to go about it, but if we can get an instance of a simulator running with geo data...

not that I'm being competitive hehe xd

it's a bad time to be changing it up cause it's christmas and stuff, but i'd say it has the potential to be quite a bit better not just in performance but in technical difficulty than the other group's work

thought we'd agreed on that

we are indeed, cause the previous system results in way way too convoluted solutions

I kind of like how it works

So are we doing this thing?

I've read a bit of OR

kthxbye

http://gamma.cs.unc.edu/RVO2/documentation/2.0/

last link of the day: official structure for this shit

not me mate

Operations research

u wot

Who's done OR? (me)

minor discovery: when they are talking about linear constraints in the paper, they actually mean the ones in linear programming, which is applying discrete optimization with matrices to a set of inequations which are the constraints

RVO2-3D is the 3D version in C++

RVO2-CS is the 2D version for C sharp

https://github.com/snape

:)

let me link you the githubs instead hold on

:(

somewhere up there xD

what's the link for that stuff again?

x2

...unfortunately

trust me I tried xD but what theyre doing is the product of many many years of reearch

aaaand unfortunately I cant modify it without understanding it

but it needs some heavy modifications to work with our stuff

we can use their code

I think we should be able to use it I can't really say how having not seen it But it's def possible

going to dinner, sec

they rely on a simulatorclass to pull i together

How do they use it then?

you cant call it

yes but its not a library at all

Yh but we can shell out to c++ that's not an issue

and while the version of c# does do 2d collisions it relies heavily on a simulator to do it

but 3d collisions are only in c++

the maths are there, we dont have to figure out how to code them

I agree!! exactly what we want

But I'll trust you that it doesn't fit for us..

Collision avoidance with spherical is exactly what we want

except we dont want a simulator cause were running our own

we need a proper structure with our own objects for vectors, lines, agents and all the rest of related shit

if we copy it its just not gonna work, I tried for almost 7 hours to adapt it

we have to use their code, more or less, but its very important that we understand how they are doing it so we can apply it to our thing

we cant use any of their functions

its a complete application for collision avoidance between circular/spherical agents

it is not a library

Does it not have functions we can just use Shame

It is a library but it's not as general enough for our use

Wtf are you looking at

I thought it was a library or something you were looking at

Exactly^

plus their application is completely different

remember we have to submit a codebase

also wrapper is way too obvious

what im doing is trying to build a mental picture of wtf they are dling

Why convert, could we not just write a wrapper that handles all this stuff in c++ and then just call it from the c#

Then trying to extend c# 2d to 3d

I'm guessing you're comparing 3d and 2d versions.

main problem is understanding wtf they are doing

Cool

yep!

Why do we have to do a conversion if a C# version exists? Is it only 2d for c#?

oli just fyi some interesting math can be found in c# version, under Agent.computeNewVlocity

bruce. kys

Will keep in mind.

useful post about minkowski arithmetic

https://hamaluik.com/posts/simple-aabb-collision-using-minkowski-difference/

i dont get how they do the 3d cone thing

I might have a look at their C++ tonight so I can get a better understanding of what they're doing.

there is something i cant figure out for the life of me

that's what I figured.

but for c# using the kdtree library

im sketching out what classes i need trying to use a structure similar to their c++ code

oh you mean from scratch yes

restarted how my friend

Have you restarted the clearpath stuff?

yes xD

I think we all will be

btw on 31st from noon onwards until 1st jan around 4pm im completely out

then go for it!

np

sorry my bad xD

shit i didnt have a clue

oh is it?

But it's online so we can collaborate

Overleaf is latex

and forget about the formatting more or less

literally just \subsection everything in

id say latex is easier right?

anyway moving on

funny how oli had to end up doing it

Oh for the report should we do it through Overleaf?

Feel free to add anything extra we should mention

Here's it transcribed from my terrible handwriting

https://docs.google.com/document/d/1w4FfFOClBUqH_6ZrEFxgI6uUJnly8F80QV5bwg-sBpo/edit?usp=sharing

It's not really. It's like a long project report.

Looks complicated

First image is last page

OK I've basically written a report outline that has everything that we have to consider (I think)

its intersect maths

yeah but to find the boundaries after

Isn't it union?

need to subdivide space into pixels and iterate or something

main problem is solving 3d cone intersection problem

yes yes that's fine

Like I said. I can still do the work but can't do any of the build stuff.

i messaged them on slack but they haven't got back to me yet

oh well xD

oh shit lel

Although I can do all the git stuff.

I can't view branches on the page

Unfortunately, I've been logged out of my microsoft account

ok I'll be able to help from 29th

probably wont push for ages though

im just gonna work on the clearpath_paul one later and tomorrow

still in master

Ok which branch is all the stuff in now?

if any of you have any bright ideas hit me up lel

im stuck, so im gonna start rebuilding shit from scratch much later tonight and tomorrow on another branch called clearpath_paul

well, that makes more than 4 hours trying to make a working 3d cone intersection boundary from what weve got

good question, given the ten days we have to get it done xD

Possible But is it FEASIBLE?

um i think that would be possible

Unfortunately no. Paul, is there some way we could get the option to browse to the test file instead of having to copy it and then type the name.

I thought I am supposed to use the log names from the log folders, since that's what was generated by test harness and server

that's correct

as in those are the tests names

Yes

I have to run tests from tests folder in visualizer

ah

did you click fly to test area

got too much shit open lel

I'll merge it into master in a bit

Need one

Setup guide :P

the drones*

but it doesn't want to load test, or at least I can't find that

sorry mb

lel

oh yeah forgot the small detail that you have to run cesium server

I get something on my screen

that works

so I ran npm install, and then node server.js

oli's works fine too

check your console

mine works perfect

http://prntscr.com/dof92r

yes

so I just open index.html?

then load a test

click that

there is a button that says zoom to test area

open it in your browser

ok, there is stuff there

alright switch to visualiser branch

oh shit that means i fucked up merge

it has only nfz.json file

go to visualiser directory from wherever you downloaded github repo

nono the version i wrote

what do I do from thre?

I get this hello world page

so I downloaded cesium

and by plug it i mean write the name into the box and click visualise

should work

take the generated log and plug it into cesium

you can indeed

You are saying I can visualise that?

I think my thing finally works. Drone are approaching and doing something

same thing for 3D code

open index.html in your browser. quite useful reference for their code

Look in Drone.cs (testharness) in the receiving messages part. I don't think it changes them.

or make the array final

so we only have to make sure the test harness doesnt magically modify them

well if we follow the clearpath thing theyre never gonna change

I don't think so. I would double check what they do on the test harness end.

do the desired velocities have to be unit vectors?

That's why we have utmvectors

I zee

Yes we needed extra information to translate from cartesian to LLA

oli is there any reason we overrode the vector class

give this man a cookie

like this

It's like U

With the union sign

and how does one union two cones

Obviously you just union the cones

alright smart guy how do we do it then

I'm gonna struggle with that unioning

Yes To union That will be hard Unioning them

also theyre doing it in 2d so ours are 3d cones which is gonna be fun to union

and if we use the nice name it sounds cooler :D

ay we have to call them somehow dont we

I watched the video and read most of the paper i think But I won't be using "velocity obstacles" it'll be more like That cone intersects that cone so they go that way

(or watch the video) (ideally both) (xD)

read the paper boys, itll tell you all you need to know

That json thing was an issue though

A setup guide will be a little tricky because for me it just werks

yeah by VO I mean the cone thing that they use, the set of all intersecting velocities to a neighbour

*

long(

well, that wasn't the main problem, that didn't take lon

Yeah that's all you should need I think But it should be like 1 minute of clicks Cant remember how tho Have you tried googling "adding Newtonsoft to visual studios"? :P

and it doesn't detect it automatically after I write using, so I just want to know a conventional way to add it

well Newtonsoft is the thing I want to add to be used by Vector.cs

Just delete it all and start again In theory you should just have to clone it into visual studio and set up the libraries it needs and it should just work

ah, it's nice to wake up early. Spent 2 hours dealing with project errors and its still before 12

velocity obstacles?

so anyway I can't find the code for velocity obstacles in the codebase from unc cause im either blind, stupid or both

Then we can all compile it

yeh we need to fix some serious shit before we submit the final codebase imo xD

So definitely get Ollie to do a setup guide

hahahaha

I get it now

Hahaha THAT'S why you get him to do it

then I see if it looks decent

I don't, I wait for oli to do it and send me the log

I guess you just pray

But how do you test it if it doesn't compile? :P

I suggest making a new one for clearpath

oh btw there is a merge somewhere that makes the cesium thing available on master so you can all pull from there cause why not

oh ive just been letting it not build all this time as long as I dont add compile errors lel

what's the right way*

well, I can make it work, I am just not sure what

When i was using the VM Oliver had to set up Newtonsoft doing some shit with visual studios before it would compile So it'd be useful anyway

well I am not sure what problems were there at the beginning and which I added

I think we should write a readme for setting it up coz I'm gonna need it when I set it up as well

I need to consult with oliver, I don't understand how this is arranged. It says package is installed, but it can't be imported, it's not in the packages, but it is also in bin/debug as a dll

oh ffs

Я не знаю

the zero clue club

makes three of us then

^ my life motto

no idea :)

does anyone know what youre supposed to do with a codebase with loads of xml comments? like how do you display it nicely in a browser and shit so we can see whats going on

but then when I try to import Newtonsoft.Json it also fails for some reason

mb it's better to use some kind of json converter rather than a specific locale

this works, but now I have to change that for drones too

good old ms products lol

http://prntscr.com/doe013

also when I tried to change that, nothig changed, cause it also decided to use some other reference to the Utils file, not the one that is updated when Utils is built

how did you fix it? im curious

lol thats weird 😂😂

right now im just trying to understand a bunch of c++ code so I can translate it lol

which fucks up the json, so the server receives null

visual studio decided to use my locale, so it converts decimals to string using comma as a separator

ayyy lmao

I almost sorted out my problem with running the server and test harness

I just also have no idea what's going on

It is pretty big So not portable for train journeys lol But I'll be back at some point in the new year

...fuck

welp

Well my laptop is dead I bought a desktop but it's in London Where am I not

any chance of you actually massively contributing or os your pc fried until after the 14th?

Oh it isn't from the library It's my bookshelf I think I found it in a charity shop lol

and you went and removed it to vandalize the library didnt you

Tbf it's currently being used to hold up the shelf above it so it's actually being useful

damn that's almost as good as an ashtray on a motorcycle

Only 11 years out of date Gonna be a pro at C#

I think I can massively contribute now

Look what I just found

Got to be slightly pessimistic. Then most things won't disappoint you if they go wrong.

first things first :p

Would be interesting to see (if we get it working) if genetic algorithms improve it at all.

yes definitely

Would be good to understand so we can explain it (at least partially)

yep in the appendix

This is the proof in the paper, correct?

but I understand the point.

their proof seems reasonable but I can't understand enough of it to tell

We can do things in parallel, you know.

by the way I looked through collision avoidance history and these guys appear to have come up with an "optimal" approach

You're both right though. On one side, yes, good practice is markable, but I agree with Bruce in that it's more important to get it working well first

I might do it myself in the next couple of days

Honestly, you could write the unit tests in a couple of hours. It would literally just be "does this drone update its acceleration/velocity/position" and "do these drones/mannedaviation collide"

Do they actually look and mark that lol Like we did it for the server so we can just say we wrote unit tests

That we're showing good software engineering practices (yes)

oh I was looking in the server lel

Rather than improve test coverage for the last week

TestHarness

Yeah idk I'll have access when I'm back in London in the new year Do we get marks for tests though? Dont we wanna make a rush to make our thing actually better

what project file is it in?

Yes

Parser for test files?

so far I've literally used empty arrays of 10 elements to make sure stuff compiles, gonna need to pull data from a test file at some point. you already wrote a parser right?

Oh. We should get someone (David or Bruce if he'll ever get access to a PC) to write unit tests for the test harness.

Cool

through piazza forum 362

35 pages seems doable

"should not be longer than 35 pages" lmao

How did you get to it?

that's it

right

https://www.doc.ic.ac.uk/lab/thirdyear/group-project/introduction.html

It's not a specific spec pdf but it's on a url somewhere

Yes and No

I think we have time - is there a spec on CATe?

also, for anybody wondering, FVO in the paper stands for Finite-time-interval velocity obstacles

Really good work. Just letting you guys know that the report's due on the 9th.

Quick update: - Added KDTree Nuget package to Visual Studio, can explain how it works to anyone unfamiliar with it. Created 3DTree in ClearPathController.cs along with some other skeletons. - Found this: http://gamma.cs.unc.edu/RVO2/ Turns out this page is from the guys who thought of the algorithm in the first place. Unfortunately their library is written for a completely different application and after looking through almost all their code I think our best bet is to translate their C++ version using 3D space into C# and adapt it to our needs in the process. - The zip file below is two of the downloads from the above page complete with SLN files for VS.

haha of course, you've done enough, relax for a bit

Not yet. I was hoping to have some sort of holiday first. I've read through the paper though.

any progress?

good news mates i sorted out my individual project server so im free to do this from now to the 14th bar family/friends/gf

Like we do with GAs

Oh a point on design. I think it would be best to create a class called ClearPath and call the method from DroneController so we can tidy up DroneController.

yh sounds good

So just get them to move in the same direction.

However they move towards each other

That

We've done thay

the one popping in? I assume you mean we have test where one just pops up in front of a drone

The only thing you would have to change the waypoints so that they move in the same direction.

Effectively the one popping in does just that.

random though, did we ever write a test where 2 drones aren't flying right at each other? like instead of going towards each other directly both going the same direction, but one is going slower and the other one rams into the back of it

7th, I think

8th probably

8th

4th January

I'll be back 7th Jan I think hbu guys?

alright good enough I guess enjoy holidays 😊

Ikr it finally arrived and I can't even use it over Christmas hahah Do my friends have spare laptops? No :P Besides I'll only be away like a week and a bit maybe Going down on the 20th Up to my dad's on the 28th and then I'll probably come back to London in the new year Then we have 2 weeks

any friends willing to give you an extra? I just don't have any xD

well fuck mate

Hence it isn't portable I can't take it on the train to my mum's lol

ah shit

No I mean it's at my flat in London, I'm not spending Christmas here :P

doesn't really need to be portable, two of us at least aren't in the UK so in-person meetings are a bit hard xD

I do now anyway

I have a desktop It's just not portable

any other solutions?

fair enuff

Lol my mum has a shitty laptop which is for her and her husband to use so not not really

we only have like... 25 days

oh btw video really does help explain what its about so watch it, then read at least some of the paper

which is probably 2-3 days from now

im working on it asap

or... ask your neighbour for a loan :')

or ask imperial for a loan

can you not use your mums computer or something

Some people are off to different countries I won't have access to a computer until like next year

But when are we gonna work on it

I know exactly 0 details about how it works but sure

anyway are you on board then?

just like the other group used APF xD

so what Im saying is we could use their technique

but if you read the paper they published they prove it works

they have no api or shared code

its an algorithm

hahaha its not a library silly

Nice So basically use a library instead of doing it ourselves Isn't that what I said at the start, libraries :P

Or you could google it

does velocity calculation to avoid collisions

For*

Is there a slightly longer tl;dr fir those who don't know what clearpath is

We're probably using it for commands

Clearpath

Is there a tl;dr

why did we not think of this

http://pointclouds.org/documentation/tutorials/kdtree_search.php

no idea as of yet im afraid xD

i have to read how they handle static obstacles

Just if the edges of the cone intersect with the noflyzone?

so how do you think this will apply to noflyzones

oh shit it uses KD trees

gonna take a few days for me to do the first and third steps lel

everyone in this group should read that, watch the video, make sure they understand and decide if they agree with the idea

Cool I'll have a read of it later

^ original research paper

need to find out how the f they parallelized

alright seems plausible

hand't noticed lol

yes you're completely right that is what it does

It's not something instant (around 1:30)

xD

pause button does not go faste

gonna download video and turn it into frames

Don't they do that in the second one (the red green yellow and blue circles)

yes but if you change the direction they're going in surely at some interval you have to redirect them again

I think if you've moved them in such a way, you don't need to do recalculation most of the time

Looks like it

I think so but not entirely sure

is it just a constant arrow pointing to a destination that is continuously modified?

I still don't understand how the trajectory is maintained to a final point

then we can say look, this works, but you have to consider lags

Of course

commubication included

id say initially we ignore that fact and build this algorithm assuming everything was perfect

We also have to take into account that we cannot instantly change velocity

why does it solve by discrtee optimization I wonder

Why not just do the edges?

I guess you can calculate the space between the drones, then figure out what innter part of the cone you can safely ignore and calculate edges only

unless you do just the edges

i think thats too resource intesive though

ummm

Could probably mostly be solved by extending the cone width

I'm interested as to what difficulties we'll have with the 1 second delay

Yeah. It would be more efficient from the looks of it as their paths all curved around

but still

obviously we'd have to say where the inspiration came from

but this is kind of more outside the box and I think it's pretty clever

good strategy

they were doing APF

This will be interesting and at least it's different from what the other group is doing

ah yes I think so pretty much

What I mean is that the underlying framework (send command as response) still remains.

well and union over geogrpahic space not trivial either hehe

its not union, i think the problem is finding closest point

but its not hard either

not that easy i think, ive been looking through all the info i can find

Meaning we can just use the current state on the server

Of cones

Because you just take the boundary of the union right?

From what I understand we can still do it the same way in terms of sending a command as a response.

That's correct

like 10-14 or whatever

if im not mistaken presentation + report is early jan

my plan is when i get to spain, spend a few days setting up a webserver for my individual project that i can leave running until summer, and then just work on this as far as I can

Fair enough. Although that wasn't the one I was talking about.

we have time I think

I vote for forgetting old approach and starting anew

smooth as fuck mate

Like I showed you last time, it becomes quite smooth.

I want to see how far I can get with the one we have

sounds good

yep if we can do this it would look great

Looks interesting. So that would replace the vanilla part of it?

this seems like a good strategy we can try and code that wouldn't take too much time if we all work on it

https://www.youtube.com/watch?v=Hc6kng5A8lQ

I'll link the video when I'm home, it's a cool idea maybe we can implement halfway before presentation

I can work after like 27th, should have lots of free time. But will need a bit of help understanding what we have now

& timing*

bad luck timing

i know bruce its not you im referring to, its not your fault mate

first couple weeks of january maybe but that's about it

this being said, im already omw out of london so if you do meet please inform me later

I would work on it but my laptop got sent back to me, after 5 weeks, still fucking broken as shit I have a gaming desktop now, beautiful thing, but it is not portable so I can't work over christmas

I'm kinda frustrated because this project is one of the best in the year and had so much potential, but 2 people in this group are permanently absent and the work we can do with the remaining 4 is not enough to come up with a product as good as the other group's

oh oli I need to show you a cool video on avoidance I found which we can maybe use

so are we not working on it further? its shit now

I'll be back in london on the 4th

well two

and then plan it

we also need to make a presentation

you have exams all day today? brutal

I've got exams all day. The only thing we really need to is write the report.

im leaving to spain tomorrow, should we meet at some point today to plan what's happening during the break?

So good news i got my laptop back Bad news is they're about as competent as a UCL student They replaced the charger which isn't the broken part So my laptop is still broken Meaning i probably wont be able to work on the group project after exams either

Yeah it was a casting issue

Too bad it's shit for the others

wait that's waaaaay better lol

Check out how smooth this is

I don't even know what bug you're talking about but yay solved

got to cast dat shit

yeah, sounds simple :D

it's because 1/10 is 0 instead of 0.1

The bug actually seems simpler

Sign the declaration my dudes

ok thanks

*3:30

yes at 4

are we meeting will today

Cool. Just got here

im late, there at 11:15

big merge inc

oh i cant push lel

i'll be there by 11

and more battery usage

it's visibly around 10 times slower

no man

genetic algorithms, my guy

cause this we can't demo

if you want to come in around 10:30

this is, again, about 20 unnecessary turns xD

not atm

are you in uni

Try this one

genetic algorithms dude

now they path in an extremely weird way

I could have sworn they were further apart

just pushed changes with pathing to visualiser branch

let me check

oli these drones are even closer than before

just gonna push now

here you go

oli if ypu're up please send over the new log or tell me where to find it, ready to test it tomorrow

yes sure

Should I just merge that in now?

yep ill be in as early as you guys want me to

Figures. Then we've got the day to trial non GA and GA.

but i said 4

hasnt answered

Oh did you specify a time for the meeting with Will?

I think I can just push that and merge then.

lmao

i cant speak english

shit

*consistent

and yes you could send it at 30 but lets keep it at 40 to be constant

okay there you go, perfect

Yeah four drones now works without spheres intersecting.

Because the yellow spheres intersecting doesn't really mean anything

You could send it at 30m

So instead of sending a command at 40m

past the yellow sphere we dont need to consider anything i think

i mean what do you mean by allowance

what allowance?

It seems to work fine with 40m

a warn command might be good but as you want tbh

although I'm not too surprised.

It made the 4 drones one timeout

BTW that made the first one work at least.

Or is there any allowance (e.g. 30m away instead of 40m)

Should I send commands as soon as yellows intersect?

yeah forgot to consider them both moving

cool

plus a few minor improvements

ill try and get the adding tests thing done

would be fantastic

exactly

so yes, the last thing to do is please make sure they are about four times more separate

that should be of little problem to change

*crash*

RIght got you now

thus they will end up at exactly the same point

safety distance is 10

flying towards each other

drones are 20 metres apart

two spheres, tangent to each other

just think about it

there is a problem

Because they're 10 < x < 20 m apart

Then there's no problem with them intersecting

no, 10

Is that correct?

So the radius of the red one is 5m

the red spheres must not intersect

which is not enough, because each one is 5 metres from a collision point

if the dot enters, they're 10m

ah yes sorry

Yes so if reds intersect they're 20m apart.

which is the 10m safety distance

red is the important one

of no use atm

Because if they intersect they're < 20m apart

no, yellow is warning

Why though? It's not the definition of the safety distance then.

they have to at most be tangent

but red spheres touching is not

yellow sphere with yellow sphere is fine

Spheres intersecting on a drone entering another's sphere?

but they definitely wont like them intersecting after last demo

if its like 0.1 seconds they wont notice

yep

there definitely is, try and quadruple the distance between them if you can

If they enter for a fraction of a second does it really matter?

Ok I'll keep working on it if there any proper issues.

only reason I kept it in is because it appears to be more compatible with the epoch thing

same problem

and it works without rounding

I round them to nearest second but I tried without too

the only thing taken into account is position of the drone from the log's location parameter

And besides, you round the times anyway.

Oh because it actually changes according to acceleration and velocity?

interpolation is correct to within a fraction of a second

no, no

okay, that's def the most important thing as of now

However that could be down to interpolation for all we know.

Almost there. They usually JUST enter each other's spheres.

im writing some of the visualiser now

did you mange to get the drones separate enough?

Therefore we won't have much to do other than demo it unless anyone can think of anything else.

I'll be merging the genetic algorithms code and pushing that visualiser stuff I've done.

shit you're right lol

oh

It's individual anyway

if nobody has written review, doing it now

Cool

very nice. just emailed will

Coordinates are actually an array of an array of an array

you have to have a property objects in the feature object.

I GOT THE FUCKING NO FLYZONES RENDERING

Oh I fixed that issue with HOVER commands in the visualiser.

Paul could you email Will for a meeting on Friday.

Yeah, genetic algorithms works really well for those 4 drones colliding for some reason.

They all passed dudes

Decided to do it so that writing to log is done every second.

Looking at writing to log in a separate thread.

Yeah that seemed to fix it

It could be a delay due to writing to logs so I might try it without writing to logs.

Ten drones at once still seems to be failing which is weird.

Any progress on the visualiser?

I agree

The lecture and stuff this year weren't helpful at all

We learned about that last year though

But generally you're right

CI was genuinely useful

We did though.

Literally

As long as you've got evidence, sure

We literally did not use any of the things he said

Well my report is just going to say that it was useless A fair assessment

You have to write bare minimum to pass anyway

Ugh wow

Is it?

It is

Is it not individual?

Just a little bit of writing and bullshit

Who's gonna do the software engineering practice thing due this week

New URL is dronesatcs.azurewebsites.net

OK that migration's done

I'll be migrating the server to the free subscription today (on our college accounts). Misha and David, you guys should look at trying to improve the hand cranked algorithms.

GL

ill do everything i can on every day not just tuesday xDD

So what I hear is you're free Tuesday

yeah scratch that i have a test on monday, interviews wednesday, coursework thursday, individual project milestone friday, gonna be an interesting week if i want to do all that plus project wish me luck boys

It's cool

should be free-er from friday latest

nice! sorry for my lack of contribution last two days, had a wedding to go to and several other things to get sorted

On the plus side, the test with 4 drones is more battery efficient than the one with 2.

Update: We have genetic algorithms for drones. Though they're a little shit at the moment.

Coming into labs in half an hour.

I didn't do much anyway. Just some method signatures.

Hahah fair Didn't see you or David or Misha so i just went home Otherwise woulda helped ya

So slept in

Was out last night

Did you come in today? I didn't see you in deep learning

Ok, just coded the skeleton of the genetic algorithm

same

So am i

In labs now

ye

around labs?

I think somewhere on campus would be more appropriate though..

Kunt

Loving the first part of that name

once you get there you can call me

Underground station Kuntsevskaya

Well How to get there

I have no idea where that is

there

еруку

So we meeting tomorrow? 11 right? Anywhere in particular?

Happy birthday!!

20

gz gz

legal drinking age in america

ayyyy happy birthday mate!! is that 21?

thanks :)

Good boy

i wasn't planning on it anyway

i won't

Something has to be sacred

According to Facebook it's Oliver's birthday today Happy birthday Oliver Don't you dare do any work it's your birthday and a Sunday!

ill be there

Cool 11 it is

Got a call to take at 10:30 but shouldn't take 30 minutes so should be free from 11 onwards Apart from the lecture 2-4

As am I.

when should we meet?

im free from 11 on monday

and you run it from there with a run config

or you download webstorm with your nice student licence

if its the first time

but you need to run npm install first

node server.js in the directory where server.js is

Just node server?

And how do that?

ayy

Ah

did you start node server

um

Me too

chrome

wut

Did you run in chrome or Firefox

Tried poll2() and got XMLhttprequest cannot load

then just write name of test, click visualise

to make sure data sources are intialised

type poll2() in console

You can just get the drone to move perpendicular to the noflyzone.

Interesting we might not have to plan for no fly zones.

How do you run it because I'm just getting a bunch of console errors.

cheers

ayy

lol

I was only looking at my branches

it

There I see

branch

switch

Dude it's not there https://iximperial.visualstudio.com/_git/Group%2026%20Altitude%20Angel?path=%2F&version=GBmaster&_a=contents

switch branch

lol

i can see it there

It's not there.

try now

wait a sec

nono

as opposed to iximperial

Did you push it to dronesatcs?

didnt i?

Paul could you push the visualiser code to the repo at some point.

right will submit in 20 mins

It's been sent

Paul could you print that checkpoint for Will.

sign on cate :)

so who's coming to meeting

or is there a math constant somewhere you can change

do you have a log with more distance between them

we're supposed to not let the ellipsoids touch

yes but thats not good enough

Well they are, are they not?

xD

"yeah theyre still avoiding each other going within 2 cm"

what are you gonna say in demo

is that you stay out of them

the whole point of safety distances

nono that cant be

They're still avoiding each other

drones are way, way too close

Having lunch rn

Alright I'll set them to the time that the Drone sent its telemetry

need a time for commands mate

That's righr

and then stops

yes

Does manned aviation move at least

literally cant see the difference

idk

How come you could render the other ones?

Weird

but im not getting any errors

changed them, no zone

must be a problem my end

No 0.001

wait you mean like 0.01 > 0.1?

Add an extra decimal

no

Did you change the longitudes?

yes

Did you translate to geojson?

xD

so idk if rendering

cant find it

Add an extra decimal point to longitudes

By quite a bit

It will be further left

Oh shit

Yeah should be just below the drones

xDD

lmao it recognized the face

"coordinates":[ { "longitude":-0.01, "latitude":51.48, "altitude":0 }, { "longitude":-0.02, "latitude":51.48, "altitude":0 }, { "longitude":-0.02, "latitude":51.47, "altitude":0 }, { "longitude":-0.01, "latitude":51.47, "altitude":0 } ]

What are its coordinates in the log?

It should be below them

Oh sorry

It should be at the top

oli where is the json "features" polygon? is it near the drones?

hoho

hoho

Of independence?

Sign the declaration guys

Ill be there soon

Same ^

Nope got lectures til 1

anyone in labs? there in 15

Guys, the Test Harness now records the logs locally.

Ok this should work now

OK I'll send this over probably in about half an hour

bedtime! ill be in labs arond 10:30

just pushed to my branch

aight

Shouldn't take long

do it tomorrow if you want

ye sure me too

Then I'm going to bed

Right fixing it now

does not move

Should do

does the manned thing move

i mean

is it supposed to work

Basically too lazy to rerun it

The way the log worked it sent the same no fly zone twice so I forgot to delete the comma

It won't

but make sure it doesnt get included

Cool

manually

Yes

i did it for you

needs to be removed

comma ending geometry

invalid json

there we go

Got it lng instead of longitude

Manned aviation is still acting weirdly

just need that

it works

no hurry mate

This is going to take a bit. My WiFi in my laptop disconnects before it does the whole thing.

0 = MOVE 1 = Hover 2 = LAND

also can you give me a quick enum with the command ids

Give me 10mins

The manned aviation should have been in the same zone.

I know what's gone wrong

Ok

That's manned aviation

Actually that makes sense

def

ye latitude 0

made sure of that

no commands get sorted out

its E-10

Unless it's a command

There shouldn't be anything there with latitude = 0

Check the logs

Wut

is that intentional

um

tyty

Here you go

then generate a log and send it over im writing the parser for that format to print commands

On it

Ok

Why would you

Don't see why not?

would make my life easier

literally just add "type":"drone" and "type":"command" prefixing the rest

Yo

you think you can add a type to everything in the log?

oi oliver

point is we can now include any geojson with one click

reading airspace

What's that

Noice

took me a fucking while

sure

We could try and catch him after Lawrence and Joni.

He didn't respond the first iteration.

no... starting to suspect hes on holiday or something

Also I'm guessing you haven't heard back from Will?

Yeah

but it should be fine if it hasnt been booked at this hour

So 402 from 2-3

havent sent it yet, can hold on

Oh ok

is that ok?

ill send message anyway

Guess I'll email them

wont let me click either

Well it says just to click on it

or call the extensio

have to send email right?

Weird I can't seem to book it

If not, head to the Sherfield room.

Check if there's a room free there.

Sure. So we'll meet them outside Huxley.

should I say that to Lawrence?

Tell you what. I'll book the Sherfield one and then if we don't find a room in Huxley we'll just go to that. L

Yeah although that doesn't include the little rooms outside labs

huxley fully booked

Sherfield room 402 is free

Imperial.ac.uk/roomsandcatering

Yes

is there any way to book a room elsewhere in the building?

Any other suggestions

Even then it's the smallest room

Shit we can do it we've can do it in half an hour

mrbs.doc.ic.ac.uk/lab

And it says it there 219A-C

me neither

Sure I dont know which one

there's some website

you don't need to be in labs to do it tho right

what rooms

If so could you book one of those rooms for tomorrow at 2pm

yup

Bruce and David are you guys still there?

Not yet

lawrence + joni

For what? Will or Lawrence?

has anybody booked a room

that is actually fair enough xD

Bugs, sure but there shouldn't be anything in active as we're supposed have everything in the iteration done by tomorrow.

so we look organized and shit

looks nice. you guys should populate a bit more, use the "active" column, add some bugs

Check it. We have people assigned now

we got a scolding from will last time

how's the board looking

We're just writing tests

Nothing much

any development on the code I should know about? about to pull

thanks bruce i know you love me

But then you might die or go permanently deaf .. Go for it

or a demolitions expert

Get a plumber tbh

hilarious

will have to wait for a spanish nurse on tuesday... lol

oh well boys

xDD

cant hear a thing from the left

2h in the hospital so they can tell me they dont have a doctor trained to deal with a blocked ear

one ear still blocked?

well that was fucking useless

well if we're still here we'll let you know

I have Japanese until 6 :(

gotta eat

well that's a bit late now

After 6?

(I'm with David)

but yeah we'll be here a while

prolly

until I die polly

How long are you going to be there?

cool 2nd row, 1 from the end on your left

Will be there before 4

so come let's write some tests

window vm seems to be working now

anyone in labs

eh effort

Yeah we would

say a directory or w.e

oh I see then yeah we'd have to agree on a way to run all the tests

and viceversa

We run our tests on their test harness and server

so if our test harness knows how to handle the tests it's ok

mind you does the format matter do they run our tests? I thought they just ran the test harness against their server

could just have files with a singleton array in each or no array at all yeah but would have to be agreed with the other team then if we already agreed to this way

Yeah. We just have to read each file.

format is the same

but its easy peasy to just not do an array

no what we said was have a specific format for one test and then have an array of those

wait what

well that's because we're lazy if it was to get used at all there'd be many many more...

We're not going to have more than 20 tests, thankfully

too late now I imagine but still dear god of bad decisions

Ask Paul or the other group. I had nothing to do with it.

or 1000 tests or more it's absurd

chaos

in 1 file what if you just want to run 1 test?

imagine 100 tests

it's hideous

all in one file? whose smart idea was that

That's how we set it up with the other group

it'd be much nicer to have them in separate files

as in atm it's just tests.json with a large file is that really a good idea?

Yes

Are we still planning on having all of our tests in one file?

ayy

welp good luck with that

but im def not showing up at the demo tomorrow with half my hearing and then taking a 4 hour flight with pressure changes xDD

i dont have a clue

how do you block your ear

but they dont do it at nhs centres

stupid ass thing to go to fucking hospital for

'got an ear block

got a flight on tuesday, got to sort this out before weekend

same thing as this morning

Why are you going to a&e??

finished task board for me btw

well... im on route to a&e at charing cross so will get to it straight after xD

so eat a bag of dicks

yeah but Mery is right next to me

You're not even doing the visualiser

no

Have you guys sorted out rendering no fly zones yet?

I'll write a few more tests today.

microsoft tm

inefficiency, incompatibility and idiosyncrasy

fair enough

absolutely. made by microsoft, unsurprising

That interface is really silly.

just changed all of mine and added some new

I got it dw

Damn I feel dumb.

I can do it too.

Ah

yes

Do you have to type your name?

just had to click the actual search button to find my id and then add that to the task

?

its fine i got it#

RIGHT i found the problem it doesnt load the users automatically

You know how each task is a small bit of it.

Did you try remaking the whole user story?

test*

yes I just tried crating a dummy one to testy

Even if you've made it yourself?

just says unassigned on all of them and I can't change it

wait do I actually have access now I'm trying to assign myself to a task and it doesn't show up

Then remake them and assign yourself to them.

You're the one that did them.

VIsualiser ones.

wait which ones do I have to remove

yes ill try tht now

lmao I meant one of the big ones cause I have an ear blocked hahaha 😂

aaaand no reply from will! yay

it's tragic, those poor heroin addicts having to go to hospital for their syringes

Oh could you delete those visualiser tasks or try to assign yourself to them?

cant believe they dont sell them at boots

had to go to hospital this morning for a fucking syringe lol

emailed him on wednesday, up to last night there was no reply, checking now

Any word back from Will?

Ok now server writes to a file even when it's deployed online.

and ill look at the safety distance spheres later

Cool

im working on parsing the log with the manned aviation

How's progress on the visualiser?

Ill join you for that

will finish up cesium tomorrow

ive been working on individual project

xDDD

lmao bruce

Doing that now

The only thing that doesn't really work is automated deployment.

Guys automated testing now works.

I think it's because I don't have the file system properly set up on the server.

So later tonight I'll have a look at that.

I can get it to work if I switch off logging.

Ok I got it kind of working.

No thanks The webapps last year was enough for me

A test one and a release (^ that one)

So what we can do is create two web apps.

http://droneatcs.azurewebsites.net/

I've published it (it's actually really easy)

OK the plan is is that I'll get a free trial today for Azure and then we can just get the credit ASAP.

okay

Paul when you're booking a checkpoint meeting with Will, could you ask him about Azure. Apparently he's managing that.

It's Knottenbelt

Apparently we have to talk to William (? probably Microsoft guy) about Azure.

Will probably leave Azure stuff until tomorrow as I've asked on slack about it.

Now drones lose connection or refuse commands according to test format.

👍👍👍

Ok here is the log with manned aviation and no flyzones

Will properly look at it and create a few more tests this weekend.

Ok guys, noflyzones are simulated on the Test Harness end.

*our

That way for the last 3 weeks, we can look at optimising are solution.

Ok the plan is that by tomorrow, we'll have a complete system with manned aviation and no fly zones.

nice nice ill take a look tomorrow

Ok guys. I've implemented manned aviation test harness side. Will look at no fly zones (which should be even easier) tomorrow.

Not till next week

guys i have a databases deadline on friday, do we have the meeting with will this week as well or is it every two weeks?

It worked the first time Tried the same computer again and it didnt work So yh idk Csg should sort it out

It depends on whether the computer's behaving or not

that's what I mean

I don't see a pattern

Depends on what? Out of the 10 or so times I've tried on different computers it only worked once lol

some of the vms work. It usually just depends.

ok

I gotta send my laptop off and get it fixed

Unless the windows vm work in labs

Yh so my laptop is fully dead it wasn't a charger or battery issue Working on it will be a problem

where id is the id of the particular plane/helicopter whatever

The url is also now api/MannedAviation/id

The first difference is that there is an extra parameter safetyDistance passed in telemetry

So for Bruce and David manned aviation telemetry is nearly the same as drone telemetry.

*three

So maybe give him each of the tree links again.

aight

ilovedrones123

what was the password for guest user?

https://iximperial.visualstudio.com/Group%2026%20Altitude%20Angel

alright

yeah that

you mean from droneatcs to iximperoal

Just mention it's the same account as last time and it's now at https://iximperial.visualstudio.com/Group%2026%20Altitude%20Angel/_backlogs/TaskBoard/Iteration%202

sure

Oh since you're here, could you email Chatley about moving repos?

c

fantasti

Will look at deploying to Azure later today. Would be good to get it working by the end of the week.

Dudes, manned aviation is now in the works. You should check out the pull request.

Yeah, we're demoing next week, though

we dont need to

Not with the Supergroup anyway it seems

Are we meeting then

not yet

did anyone ask ms for the extra devices?

been super busy applying around to see who'll take me, been out of it for a day, will hopefully finish all the important cesium bits by tomorrow

OK now server generates commands when considering noFlyZones.

Ok added noFlyZones to server and checks for collisions. Doesn't generate commands based on them yet.

Oh to do that you could check if the line intersects with any of the sides of the polygon

Because the point where a drone might end up is outside the polygon but the path it takes goes through it

Actually we should consider intersection with a line and a polygon

oh we used something very similar for our geojson group project last year lol

Useful for no fly zones (as they appear to have no altitude upper or lower bound)

I don't know useful this will be to anyone else but here's an article for checking whether a drone is within a polygon. http://alienryderflex.com/polygon/

Awesome

just to let you guys know, the visualiser will def be ready to demo on wed

Oh I've added some tasks for this iteration. https://iximperial.visualstudio.com/Group%2026%20Altitude%20Angel/_backlogs/taskboard/Iteration%202?_a=requirements

fair enough. I'll ask

Probably will be telemetry like drones but better to be safe than sorry

Well how else would you do it on the server end?

Should ask

That sounds like telemetry to me

There's defo more than one way to do it

Who knows They plan flight paths in advance so there could be some kind of feed with updates coming in as stuff changes

I mean, how else would the server know the location of manned aviation

Seems like a question for Altitude Angel

We'd need to change quite a bit i think but can't say for sure without the code are we expecting telemetry from manned aviation in real life version?

Should I assume you guys would be sending telemetry from manned aviation?

Well. It's not my problem as I'm doing it server side.

I assumed they would be fairly simple (only moving in a single direction)

Yep But we'd have to then limit their direction and stuff E.g for now we've assumed you can accelerate in any direction

Because with manned aviation you could simulate them like drones. Even sending telemetry data to the server.

Are you talking about manned aviation or no fly zones?

If that logic is done it can be used for both sides

The hideous 4d array or whatever it is

Of course

Yeah But i meant like storing it in our "world"

Obviously manned aviation is treated almost like a giant drone.

I'll be looking at server side. Although I don't think we know how it's represented. geojson?

Yes

Cool So test harness side after that's done we just need to implement parsing it from the file and it's done

Master is most up to date

Coz then the work would help both sides at the same time right

We have merged, yes

That representation is the same for the test harness and server right so have we merged?

That's fair enough. I'm probably going to head home around lunch anyway. Looking at dealing with manned aviation + no fly zones.

Most likely But to try and sort out some robotics first since that's due Wednesday

Is anyone coming into labs today?

I've been having the same problem😬

Just to let you know, my laptop charger and/or battery has passed away so I can't really do any work on the project unless they get the vms on the lab machines going or I use someone else's laptop

figuratively, darling

at this hour? probably at home

xDDD

and where's the group

where's the leadership

tbh I blame the group leader

i dont see anything happening whatsoever

oh well I don't see it happening at any point other than just before we show him the board lol

We'll also have to deploy to azure

yes, we should do it properly xD

He specifically complained about unassigned tasks

but that won't happen

or we could do it properly

like todo, done or w.e

oh ok if we can create tasks though can have them unassigned and just in the right place

*our

I can't assign someone else a task, for example

Are permissions are still shit.

but the board we're not using at all

well idk what state the CI is in and all that

which is a fair point because we're not

rest was ok

he said were not using the team stuff enough

need to work on the board

what did he say

how did the meeting go

Ok so I'm not needed right

Im going too

I'm going

Labs closed so David and I went home

Who's going to the meeting then?

yeah i cant count

it was element 22 i had made a a mistake

Told you to get rid of the commands

cant believe it

fucks sake

Ah

and is trying to read it

its a top-level element in the log

its the fucking command

omg

Or maybe ignore item 20?

Just try with 19 then

i did try yeah

Weird. Have you tried deleting that object or is it the same no matter what?

it says the object is undefined right when it reaches number 20

*in particular

Why that one I particular?

i dont

Oh I understand now.

That means nothing to me

k it's object number 20 in the log that's the problem

That's really weird. Keep at it.

if it was misspelled it wouldnt return it on console log

Damn

i just wrote it

no

I assumed you just copy and pasted it

Well telemetry is misspelled

yet i console log it and its fine

nd the location object is undefined when i access

i want to access telemtry[i].location.longitude

no, im not even checking the commands

It could be because commands and locations are different types?

welp, that's animations gone xD

I don't know. Sounds like a js thing

put it through an online validator just in case and its all green

its valid alright

ive been checking for 30 mins xD

Double check if it is valid json

yes, I'm doing that

Not just an array

Remember it's an object that contains an array

yes

Is this for parsing?

but when i log it its there

im getting an undefined error on an array

What's up in particular? I might be able to help

It ends at 4 but yeah

after lecture i mean

oli u around?

We'll be showing the visualiser and are whole system working.

I'll be there

I don't think so

well given last time it was just 3 of us

i thought everyone had to go

that's at like 4 right

so whos meeting the supervisor and what are we showing him

Updated log

here you are

will do

the supervisor

the one he needs to sign

has anyone printed out the sheet

Getting lunch but I will send one ASAP.

oliver can you send the logfile we were looking at the other day over please

✅

Signed

Cool signed

done

DAVID AND MERY

and we don't have people signing late

so we're not chasing people up at 6pm tomorrow lol

give everyone time to sign

btw create the declaration soon as possible

yep

ill get a demo working and then worry about parsing them

just in case parsing

which i dont care about at the moment

There are commands in there too.

yes exactly

of json objects

The logs are represented by an object called "telemetry" which is an array

ill check it out dw

its an array of json objects right?

Rather, they do

Yes, it does

latitude longitdue

lat lng

lating?

and get the id when you click it

and I can make the drones clickable

the logs have latlng in them yeah? literally all I need for the demo at this point

If you need logs we would generate one as an example.

Would it just be various visualising examples.

um yeah sure I can make a demo work for that time

Will you have anything to demo?

Awesome

yes will agreed to meet at 4

We can test this all in the morning.

Just pushed that print.

No haven't merged yet.

have you merged or anything yet we're still working out some kinks so there'll be some changes coming

yes

or

what to the screen

printing?

So that they're printing on the test harness end.

There's a slight fix I'll do with the commands.

Actually it would work, it would just be adjusting the acceleration constantly

I support it fully in its decisions

it might it can be whatever it wants

as the magnitude will never be exactly 0

For the diff.getMagnitude(), that's a double so I would put it as <= 0.01 or something.

We should not have problems as it's a separate folder.

ok

I'll let you do that mr pull Request

That way you don't have to worry about multiple repos to demo it.

I would merge if I were you.

cool

makes shit neater

and you can overload operators so vectors have - and + now

I did that server side.

Yeah

just call getMagnitude

instead of creating a new vector every time and doing euclidDistance

btw we added some utility stuff

then it'll just take off again and fly to the next waypoint tho I'll leave it for now

It will then fly to that waypoint

Add a new waypoint to the front at (long, lat, 0)

Now, I got it.

Yeah for landing I would suppose negative altitude.

Awesome

That's right

so beautiful

yeah so we're done then with the target velocity stuff in place it stops pretty much exactly where it needs to and whatnot and commands are summarised by: switch(c.command) { case CommandName.HOVER: targetVelocity.setAltitude(0); targetVelocity.setLatitude(0); targetVelocity.setLongitude(0); break; case CommandName.MOVE: targetVelocity = c.targetVelocity; break; case CommandName.LAND: default: throw new NotImplementedException(); }

Most of the time you do not want to land

It will be more for load balancing in the future

Yeah, I haven't done land either because there's never a situation where I would issue it.

not land

yeah I meant hover and move

so we're gonna leave that for now

Hover and move

but land will be a bit trickier

we've done the land and move commands

have you done land on the server side?

cool

yeah but commands can just set the target velocity and then the drone will take care of it, makes life easier then when the command is done it'll resume waypoint stuff and the target velocity is set in there

Funny typo

*stone

Although they're only dealing with one target velocity at a time.

2 birds with one zone

Exactly

that was the idea commands would need this stuff to work properly

cool

it should decelerate

yeah exactly, so if you're far enough away from the waypoint, targetvelocity is the max otherwise it's set accordingly and if you're close

In that case responding to commands is a slight extension to that as commands can have a target velocity

and slows down to reach it

do you mean so it properly stops at a destination?

*test data

But we never give it a target velocity in the test harness

Oh ok

and then you can stop and accelerate according to distances and stuff

we;re making it so you can have a target velocity

nah it sucks

isnt moving to their destination enough?

In what way?

nah it's fine for the moment we're making the drones navigation smarter

We have to merge the two anyway

Maybe merge with master and create a new branch.

No they are not.

are commands sent anyway atm on the test harness branch

that way commands are never sent.

do one where they are moving close together but not towards each other.

any thing >= 0 longitude will be in another zone

yes

so change the second one to be -0.0something or

what do I change? the longitude on both is -0.01 for the first waypoint and 0 for the second

maybe moving into each others starting positions

so do a test with two heading towards each other

Just a reminder that 0.001 in lat and long ~50m

Aren't you looking at how a command is processed.

this isn't very helpful

but change it

yeah we need some actual tests for knowing whether stuff works

in that case yes

the first one is meant to time out?

keep longitude < 0 and it should work

I think the first one is supposed to as they're in two different zones

I don't remember

and the second collide?

is the first test meant to time out?

we're in labs working on it now but we're making things more sophisticated, because it sucks atm

we were messing with you

What is the status on the meeting? Paul, has Will got back?

defo tomorrow

but plz

Probably

cool so we can do it tomorrow before the meeting

It would literally take maybe 30min max

But after that it's done.

Well you have to generalise a bit of the code.

oh good so it's done then, we don't need to do anything

I'm not asking for much. In fact, the code to do it is already in drone as part of turnTowardsWaypoint.

maybe

maybe you should program us to respond to commands instead

*to do that

As in you guys are supposed to that today

good idea, thanks for offering

what, you are?

Getting the test harness drones to respond to commands.

guys...

what progress? I'm at home, dunno where David is

So how's progress on the test harness?

I dunno. Does 4 sound good to everyone else?

yeh what time

Paul could you send an email to Will ASAP about a checkpoint meeting on Friday?

How did you optimise it?

omw

sanic

Optimisation of neighbourhoods is also done.

couldn't agree more

I'm also making neighbourhoods more efficient. A massive array is a really shitty idea.

But it will work in most situations

It's really basic atm

looking forwards to seeing that

Ok guys. Now the server generates commands and sends them back in the response.

Fair enough

I'll be in around 3 to check out what you've done and update you guys on Cesium, choosing individul project now

Ok

David and i are in labs But doing robotics

Anyone in or coming to labs?

Now looking at what command should be recommended in what situation.

Sorted neighbourhoods entirely. If anyone wants to look over it there's a pull request.

*applause*

there we go

wish I had a "terrific" gif

My dudes. I''ve sorted out neighbourhoods. The only thing I've got left to do is write unit tests which I'll leave until tomorrow.

mate youre great xDD

working on cleaning it up into something understandable so we can test

there is an absolute mess of console logs going on in their copy

lel

til vs has a browser built in

we're looking at managing drones

we've nearly got lla done

now to see how the fuck you put a drone on it

finally got it working

small difference cxD

Where you at

I didn't see you but maybe I'm blind

Are you still in quiet labs?

We're pretty close

We're just working on converting LLA to cartesian

But there's someone here now who can collect it for me so I'll come in

I would come in but I'm waiting on a delivery

Any progress on changing the test harness?

Quiet labs

rather than absolute

It's more that x and y are relative

Yes I'm here

yes I agree, only methods I found are simple conversions

Anyone tbh

You still in labs?

It will only really "work" in one zone at the moment.

OK, it's done on the server end.

Paul have you managed to find a method in Proj'.net that converts lat and long to zone number?

Ok looking at library stuff on the server end.

Because the second parameter takes any name

Weird

In DroneController.cs of the method called PUT

what signature?

Apparently in the signature the first parameter has to be called id

Oh I figured out the reason why the PUT wasn't working.

Alright. In labs now till probably 6

ill let you guys take care of the library for now

will check it out at home, going there now

however the cross-origin policies on chrome are fucking me again

their json is literally 9 lines

k should be easy

well i forgot my slack password but mi on it yes xD

Maybe test that cesium stuff out?

I found a problem now that the server doesn't take PUT requests so I'll be checking that out

oh just noticed lawrence sent over the rehash of cesium

Other than looking over the library and migrating to the new Vsts no

ill get on that as soon as i get to my laptop then. you done anything this weekend?

Exactly

alright so the library shit

I suppose it's LLA conversion atm.

im available to work on anything you guys need today. still waiting on microsoft's cesium api, but i can help on whatever

fun times xD

talked to like 15 people, nobody finished

50 minutes for 4 hard exercses

just finished test

And lunch

OK I'll be in by around 1 or 2 as I have to get my laptop.

I need to sort some stuff out anyway.

Cool

after that I'm there

I have a test today at 11

Is anyone coming in to labs today? I'll be available from ~11

and that's the proj.net example

http://www.sharpgis.net/post/2007/05/05/Spatial-references2c-coordinate-systems2c-projections2c-datums2c-ellipsoids-e28093-confusing

for david with love

Proj.NET performs point-to-point coordinate conversions between geodetic coordinate systems https://projnet.codeplex.com/ http://blogs.u2u.be/diederik/post/2010/01/01/Converting-Spatial-Coordinates-with-ProjNET.aspx

At a desk

Yeah

Main labs?

Yeah but where in labs?

Huxley

Labs

Where?

David and I are in labs

I have a lecture but after that I'm skipping tutorial so I'm here

"maybe" xDD

Why?

Maybe.

any of you here after 5?

It's easy to rename

yes I saw

not an issue but still

but we're 28

like the repository they set up is group 26

btw did anyone else see that altitude angel thing we're the wrong group

Maybe this http://stackoverflow.com/questions/1185408/converting-from-longitude-latitude-to-cartesian-coordinates

and found nothing

other team has looked at that already

Or origin is centre of the earth

So we just have to find a library to convert LLA to cartesian and vice versa

Not really no

or wait does it?

okay, but wouldnt stuff change with curvature of earth

So I was thinking 0 0 0 cartesian is just 0° 0° at sea level

okay

We're only using cartesian

We're not referencing LLA AT ALL

so thats not being done atm right?

yes exactly thats the problem

What we can do is convert LLA to cartesian and work with it, only switching to LLA for sending telemetry

wait so how are we doing x y z

No we're not considering LLA atm

but you just said its not cartesian xDD

By cartesian

so how are we doing distance calculations?

We're not atm

how are we doing projectons onto cartesian space from LLA?

:D

yeah, that's what I meant

Although it checks up to 2 seconds ahead

It checks every time the state is updated

hm

Not at the moment no

yeah, but I think it checks not the current state, bit also whether the collision has occured in between checks?

But it only checks for the Drone in question

Which checks on every time data is sent to the server

yeah

I think

We're talking about David's code in the server

if I understand correctly

mb there is no collision at 0 and no collision at 1.0s, but it occured at 0.85s and David's code is to check for that

Since we have up to date info with the test harness.

We check every update

Ok

not sure how often we check, I used 0.5s as an example

that's for david's part

Because for the server we check every time telemetry comes in.

Are we talking about test harness or server?

point is that if we check for collisions every 0.5 seconds for example, to detect whether the collisions occured in between these two checks

We do so that.

ty for info anyway

i thought we were gonna do straight lines between drones for that

Its the intersection of 2 spheres in 3d space solved to find the time of intersection, there isnt really a quicker way to do it.

well yeah its the group chat he'll read it and answer hopefully xD

Like I said, you should ask David about it

Yeah

same with matrix multiplications

programming-wise I mean

yeah I was wondering, cause there must be a much faster way to do it with the same math

He's the one that knows the maths

Basically it's a massive quadratic

he'll read it too

Ask david

oli can you make me a summary of what checkCollision in DroneController.cs is doing mathematically

Yeah i mean a more fancy one😃

Sure. Shouldn't take long

We're probably sticking with it.

how about we do this weekend

so lazy to do that

Oh yeah, we'll have to migrate to AA's one at some point.

already on vsts https://dronesatcs.visualstudio.com/Drone%20ATCS/_backlogs/TaskBoard/Sprint%202?_a=requirements

Guys can i suggest making a Trello (or similar platform) board so that we can keep track of what needs to be done, what's been done and what is being built at the moment so that everyone is aware and everything is more organized?

and are we sticking with that model? I think that's good

At the moment test harness generates them.

so what's the final verdict on uids?

sorry went to p/o

damn

Don't think so as I dont think c# uses any ponters

Don't know

can you do function pointers in C#? having dedicated functions for the events would be cool

So at the moment just convert to objects and stick them in a list associated with the drone.

The only thing you have to consider are actions which only concern special events (losing connection, shutting down, etc.)

I've changed most of it now

just what stuff we read in and how we read it in?

that aside what do we need to change in the test harness?

and you have to store the mappings

that could be another internal id but again that's more work

so we need to map the id to something

we need to keep track of id though we can't do it by ip address or anything like what if a drone goes offline for 5 seconds and appears somewhere we need to update our system and say oh that drone is over here now

great job

Now will consider time limits

my point with all of this is that we need to make it as easy as we can for ourselves and in my opinion one step towards it is to forget potentially bad info from the drone and track it with our own ID

Alright I've changed the test harness so it will process the new format properly other than the extra stuff.

we will probably switch to that at some point but it makes no sense for just a few drones

remember you can just use C#'s GUID

and just use our own ID system

I've been working on the test harness and added a counter for uid which increments each time

I'd say forget what the drone is saying except telemetry

in that case youre creating 3 or 4 edge cases more we have to test

because of course uid's will be paired with the drone's endpoint

we just need to check it's not stupid

no if they're sending to us there'll be a convention so they'll have the same kind of thing but yes we can assign uids internally but that's more work for us, if the drone is sending a uid I see no reason not to use it

what do you think?

it makes more sense like that to me because you avoid any problems like bad UID, just handle it internally

we assign an internal representation to each drone we receive from

but they dont though, because manufacturers will have different ID conventions

that includes sending bad uids

we have to be sure malicious drones can't damage the system right

well I mean in the real situation yeah they come from the drone

well number or hex string or whatever

if the UIDs came on the drone then yes you're right, but we have to make sure the number is safe

well, considering it's our job to assign the UIDs that situation should never happrn

or something super big

like if the server can handle -1 being a uid

yeah unless you want to test uids

I would say so yes

It's the only not dumb way of doing.

Test harness keeps track of uids then.

Ok I understand.

spec*

abd if we assume initial velocity 0 for all then we dont need it in the test spe either

be that the test framework or the server

they are assigned by whatever processes the file

what I mean is that the json test spec itself doesnt need uids at all

sorry i havemt explained myself properly

can't immediately be moving

But if uids are assigned by the server how does the test harness know what they are?

is this in the test harness? it'll start at 0

I would rather assume it's 0 velocity

ew

just assume max and simplify thibgs

and the initial velocity is just gonna be 5 isnt it?

Ew

drones are not missing a uid as thats assigned by the server upon receipt

oh ok

the actions are there to indicate deviant behaviour

wypoints are whatever you want them to be, from start and end to just a trajectory

Drones maybe should have an initial velocity

Oh yeah. Just realised, Drones are missing a uid

Or are waypoints now only start and end?

I don't understand why there are waypoints and move actions?

So David and Bruce, could you rewrite the TestHarness ASAP so it's compliant with this spec.

see if there's anything we forgot

check it, modify it

standard json test format as far as jan and me agreed

I am

okay, so who's gonna be here for call?

Yeah it does look bad. I have been relying on the division of work that we decided at the beginning- paul is obviously very busy and the part that we were gonna work on is pretty much resolved so please guys let me know when you're working so i can come and join you and get involved in the other parts as well

okay, that's great, was just curious

Well David and Misha worked with us at times. We were just the ones logged in.

okay, I'm not one to talk here as I've been super busy and haven't written a single line yet, but it's sad to see only oliver and bruce on the repo history

Np

thanks mate

fair enough

Some library that lets you do cool shit

aaand System.Linq?

Interface

Interfacw

is it just an iterator or am i mistaken

sorry whats the difference between dronerepository and idronerepository?

Only some of the .cs files

Nope

please tell me it wasnt you guys who wrote the assembly lines in there too

right

Logpath is a variable in web.config

that

System.Configuration.ConfigurationManager.AppSettings["LogPath"].ToString()

not that

oops

There are currently no undergraduate applications available.

what does this do?

okay, let's just not worry about it now xD

They're automatically generated

precisely

I don't know

???

under the content folder

okay, why is bootstrap there?

Part of asp applications

just part of vs?

what on earth are all the .asax and such files

That's what checkCollision does.

We're doing straight line and trajectories.

Of what? Most of that I borrowed from David's code.

we're sticking to a straight line between drones to control the distance between them for now yeah?

I see u did the math

Vector.cs

were is getLatitude() defined?

Test harness also has a lot

Of the server yes.

dronecontroller.cs is the main chunk I'm guessing

wow, you've done quite a bit mate

25 minutes on this shit xD

i dont know what the problemwas i just removed all 22 microsoft cookies and logged in and it worked

fucking finally

Try logging out and logging in again?

Must be a bug

Yep

welp, i cant log in xD

paul.benn@hotmail.com yes?

Hotmail

Oh did you spell it dronesatcs?

what email account does it list under my name?

You're authorised here

its probable im doing something wrong but i cant really see what

for some reason

401 unauthorized on my account

Define locked out

anybody know why I've been locked out of the drones vsts repository xD

Feel free to add anything I've missed

Here's a catch up for those who will be at the call tomorrow but don't know what we've done https://docs.google.com/document/d/1kRF8-CfZpzG3Ea-rsIJ1cRdTnay__Ju0Jkpg4ShxB2o/edit?usp=sharing

I don't know how we'll be doing it at the moment.

Obviously this will be moved eventually to a collection of problem "neighbourhoods"

I'm thinking maybe each drone as a list of drones it could possibly collide with.

Although it doesn't update anything at the moment.

The collision detection at the moment checks every time the particular drone is updated.

Alright, I have drone state representation and collision detection sorted server side.

Exactly

I'd say we concentrate on getting something really slow that works and then trying our best to refactor and speed it up

So we can do basic detection atm (spheres and current trajectory)

data*

well, that's one of the things we're gonna have to heavily optimize to consider as little as possible daa

Good point

yh also need to think about time since the update and all the places the drone COULD be

Which is my job

dang we need to think of a way not to store all those distances between them and just the problematic ones

Just euclidean dostance

We already have their locations

so we do it by measuring a virtual line between them?

If the distance is < 2*safety distance for then they are in each others' spheres.

bruce pls

within our group of 6? maybe not

yeah, I'm sure somebody does

Can't we just use the radius.

does anybody know the math for interecting spheres with cartesian coordinates?

I'll work on collision and safety distance detection.

I'd say since I got something that turned out to be pointless thanks to Microsoft eventually giving the code to us I'm able to help anyone who needs it

So how about you guys get the drones in the test harness to respond to commands.

& yeah I know fuck all about server stuff

until test harness is done I see no reason to change it then we can properly design algorithms and such

I suppose you guys wouldn't know the server code as much as me.

Previous iteration

I thought we only agreed on it for this iteration?

makes life easier

so we should stick with them

nope but we assigned those roles

Does it have to be?

test harness is David and me

yeah but you're server team lol

That's kind of the point of branches and that

Time

Tome

We can do two things at the same tome

argument goes both ways

no use following commands if they're not sent tho

No use having commands sent if the test harness won't respond.

which is the opposite?..

Yes

you said you were gonna work on following commands in the test harness though?

Exactly.

then have it send commands / warnings

shouldn't we do that then? like make the server have some smarts

If anyone wants to work on collision detection (both path and safety distance) go ahead in a new branch.

right

I think only misha and mery can't until we get that credit.

Well most people can.

vsts is set up so we can all edit everything right?

I'm in lectures and busy at least until tomorrow night with the course I'm doing, but I'll be on here all the time to help out

oh I see what you mean

I fixed the problem of assignment though.

Storyboard for iteration 2 not set up.

Having the service detect collisions.

I thought we had 2 weeks

because we probably should try to do scrum at least vaguely

wait what do you mean now

also have we sorted out the storyboard yet

it's the second iteration now right what was our supposed goal for this one?

I won't be in labs today but I'll be working on the test harness at home so that drones respond to commands.

I suppose converting to latitude longitude too.

and the enviroment

in terms of test harness, the only things we need to consider are near misses (don't hit but distance within safety distance?)

fantastic

Just confirmed that it works with multiple way points. The last way point is the goal.

It was actually really easy since acceleration changes instantly

Noice

Although they don't stop once they hit the last target, they basically circle it in a tight circle. Experimenting with multiple way points atm.

Good news everyone. Drones now accelerate in the direction of their target.

David did u send an email

GODDAMMIT DAVID

David you should send an email too because you didn't sign until today either

So don't worry

The same thing happened last time and they said as long as the group submitted on time it shouldn't be a problem, ill contact Maria or Mark again.

Guys i really thought i had signed it, idk why it didn't go through

well, it was good while it lasted

We all die

what happens if a group member doesnt sign?

ffs

GODDAMIT MERY

Mery... Sign cate

Good idea

will definitely check it out tomorrow, going to play social life a bit

If any of you want to look at it it's the build definition called Test Harness definition.

It's probably that the msbuild arguments are wrong.

hmm curious

I don't think it's that so much that it never is deployed in the first place.

or a command line flag with an array of open ports? must be a way

Yeah through vsts

maybe you need x11 forwarding?

you are trying this on the microsoft build server yeah?

Yeah I seem to be having no luck atm. The port never seems to be on.

So it looks like it builds but then it doesn't run it.

I've run two different builds (one that builds the server and one that doesn't) in both cases the port is not listening.

Ok, what seems to be going on is that it builds does not run the service, meaning that the port isn't open.

http://stackoverflow.com/questions/1116175/running-asp-net-on-a-linux-based-server

Apparently it might work but it might be limited

let me know if you can, im curious

My guess due to running the service on it. Although I can't confirm.

why would it have to be windows sorry?

Well it would have to be Windows but I suppose that would work.

that way we have them 24/7

um if its just an ubuntu server we could just pay like 2 pounds each and get 3 of em somewhere

So I could potentially just run it on my machine, but it would have to be on at all times to be useful.

2. Or build the service on a different server with the test harness on the build machine

1. Use some other server I can change the port access on for the build server

There are a couple of options.

I'm not too sure if the service isn't running or you can't send on the particular port on build machines.

I'm trying to diagnose the problem atm

can you re-explain that paragraph for somebody more stupid please

surely CI wouldn't make the tests impossible

?

I don't understand the thing about CI

So we'll have to wait until we have Azure credit before trying that.

Tried a build definition that almost works. The problem being is that I can't send telemetry to the service so it fails.

Looking at making test harness run with CI. The only thing we wouldn't be able to do is visualise the tests so we don't have to shut down and restart the server as the logs aren't important.

doublechecked

yep

And it's with the pdf I sent, right? ;)

created on cate, should be fine

shut up xD

Danger Zone!

Paul is coming

coming

Here's the signed thing for declaration

so we can all sign it whenever

Paul can you create the declaration?

the gifs on this group are too hilarious

xD*

and only for like a few days as well... considering just paying for my ownxS

ict**

ah and of course doc vm but kind of useless given outside access is only enabled by going through a lot of red tape with icy

Im very busy until I finish an external course im doing, which will be the coming week... after that im full drones

idk what mery's done tho

I figured out how to put entities on the map, including an actual 3d model of a drone, how to make animations using the built in cesium clock and how to properly import json data without having to use a parser. however, I havent made my own example as it's kinda pointless when there are so many online and I dont have a dataset yet... I have many questions though, the main one regarding how on earth you stack entities to consider them as one. in other words, no

What's the state of the visualiser? Anything to demo?

will hasn't responded

good luck

I have an interview now but I'll check later! :)

Ok

Not a call to the other group That'd be a bit pointless

And Microsoft

Yeah with the supergroup

also, we have a call at 14:30 right?

Good question, Has he replied to the email Paul?

Are we meeting at 4?

*at the current point

A suggestion for getting it to stop at the waypoint. Check the distance between the drone and the waypoint. If it's within stopping distance and the current point in time, slow down.

4pm?

yes, what time would you like to meet

um

Paul could you email Will at some point to arrange a meeting tomorrow for the checkpoint.

We've got the test harness (sort of) working. It checks for test failures but that's it.

I may not make it to the Microsoft call

by the way guys I have an interview on Friday morning, out of town

Yes

mery, misha, sign

at least we dont have to learn something else lol

yeah

it does use git right?

go ahead then

they work fine

i just did

sure. I would double check the other links

ready to submit?

it does indeed

check it and it should show the board

yeah

that's the tracking service yeah?

cool

yep just changing now

Paul, could you change the tracking URL to this one? https://dronesatcs.visualstudio.com/Drone%20ATCS/_backlogs/taskboard/Sprint%201?_a=requirements

Are you going to submit the file, Paul?

Ok so I've merged my branch in

gedit indeed

Use gedit Beautiful

Ew sublime

in sublime its perfect for me

lol

It runs as one line

It reads really weirdly in notepad

see if everyone thinks that's okay

ok

just gonna provide a short paragraph on why VSTS

Great. Now we have to create a scrum board

yes, all 3 links work correctly

sorry no coffee today

alright

oh i see what you mean lmao

urls in file

you know the whole point of that file

drones in the file

yep I'm using https://dronesatcs.visualstudio.com/Drone%20ATCS

urls also work

should I be checking anyhitng else?

the guest account logs in correctly

I notice a slight lack of participation from 3 group members here btw xD

So just sign in as a microsoft account

right I see it

it's in the file

I assume it's on the VSTS website right? what password should I use?

cheers

of course

Could you test if that guest account works?

thanks mate

Here it is again

it has been sent

sent*

Lol

maybe "flesh it up" like will says xDD

I'll take a quick look and proofread it

has the tools file been sont on here?

right its done

hallo! will create in a sec

Fair enough

Oh it matters lol we had a whole issue with it during pintos last year

But yeah he must have if I can't find him listed.

it usually doesn't matter in my experience

Group leader has to do the declaration i believe

Oh the reason I asked about the declaration is because I tried it myself and couldn't find Paul in it.

Might as well

Lol testing

forgot to write tests so getting lunch and then doing that

Ok I've basically finished the server (logs data and that's it)

Paul have you create a declaration for the deadline today?

https://www.asp.net/web-api/overview/testing-and-debugging/unit-testing-with-aspnet-web-api

Two pretty good links on WebAPI (for REST stuff) https://www.asp.net/web-api/overview/older-versions/creating-a-web-api-that-supports-crud-operations

Meh I'll look later

There might be something in settings.

I'm not sure. Maybe your own username? I don't know.

doesn't seem to work havent been able to clone it

tried the email, and the first part of the email

what's the username? tries to clone from 'https://username@dronesatcs.visualstudio.com'

It uses git. I guess you clone from there.

it uses git right, but how do we set it up with the project Oliver set up

How do we use the version control thing?

yeah well I guess if we represent "hazards" in general as no fly zones then drones could also be no fly zones so we should be able to add and remove them dynamicallly

Represented it is like representing giant stationary drones that don't do anything.

It's not hard. It's more like, can a person add a no-fly zone at any time or is the environment is set.

oh ok, yeah that's a tricky one

We were just considering a case for later (How the environment is represented)

did you discuss much after I left? I saw Misha, Oliver and David stayed

Labs Now

Yep

is there a meeting

PLS

Yeah, you just get one of them to...https://www.youtube.com/watch?v=cEuU64Zt4B0

and we have geojson

viewer.dataSources.add(Cesium.GeoJsonDataSource.load(

yes, but they can't avoid each other if they don't know they're gonna crash or know but don't react

but still all these edge cases will be fucking annoying

I'd say a 10 metre sphere is already pretty big so I agree

Could they not avoid each other without stopping?

how about if two drones are going at the same speed towards each other and one stops responding?

So maybe 10m? Nice round figure.

I would add an extra second to account that for commands

we have to add transmission delay... and what else?

thus 6.25 is the absolute minimum horizontal and vertical safety distace to another drone

Uh huh

sooo if a drone has a max speed of 5 metres per second and an acceleration of -2 metres per second squared, it takes it 2.5 seconds and 6.25 metres to come to a complete stop

Right I'm going home. Will be back for the meeting.

holy shit they have 3d terrain worldwide

lmao

Fucking schweet

just figured out how to do animations in Cesium

news*

good new

AH

just shared a Google Doc with all of you for the questions via Imperial email just in case we think of anything else

Oh here's that file we need for wednesday

why use cs when you can use your own useless format amirite microsoft

oh yeah it uses the fucking solution format doesnt it

And the initial build failed because there are no VS .sln files yet.

And one us can use it of course

Otherwise, the guest account cannot see build information.

noted

Ok I've got a question for Altitude Angel (or maybe Microsoft). Would we be able to get credit on VSTS as it's only free for teams <= 5

I just sent an email to CSG requesting outside access for the VM with cesium on it

great!

and I'm in

Ok new url is https://dronesatcs.visualstudio.com/ You should get invite links in your microsoft accounts

Come to think about it. That will be easier anyway as integrating with Azure would be easier since we're using personal accounts with that.

As I think it's the only way to add a "guest" email so that other people outside our group can view it.

Ok just so you guys know I might have to re add you with your microsoft accounts that are not uni ones.

cool thanks man

Ok changed it https://droneatcs.visualstudio.com/

I'll see if I can change it later. Getting lunch now.

Kinda dumb that it's like that

thanks mate

got it!!

yep

oh it's on his page?

that's the one it's on

go to the one Oliver sent me earlier

are you on your page or Olivers?

(?)

however I only have "My First Project" listed

yes, I found "Browse"

it's more of a link rather than a button

there's a new and a browse or something like that

it should just be on the page he sent, under the bit about projects

do you guys have to invite me or...?

I acan't find the project

where is the "browse" button?

We'll only have to create a "guest" user so that others can view it.

So basically we have our thing for Wednesday set up.

That's what I'm saying.

If we have visual studio sln files and test files it will build those.

It builds nothing at the moment.

what on earth does it build?

At the moment it just builds and runs tests which is really the only thing we need atm.

I'm assuming it's hosted on MS's servers.

Ok I added a build def. We won't be able to see how it works until we add visual studio code.

must be distinction between ones you created

now we need to learn how to use this

c-ool

but yeah found it

but yours I had to click browse

ok

on my site it had my "first project" listed

yep found it, had to click browse

Because I have the project here.

what page are you on now?

I guess we don't have anything yet so nothing is showing up

ah I see everyone under users

but I don't see anything useful

it loaded which is an improvement

Oh I think I get it now. Try otn14.visualstudio.com

but yeah nothing

I'm at https://bk514.visualstudio.com/ which is what I set my username as

Go to [yourname].visualstudio.com

oh right

so nope

I have an empty project like "MyFirstPorject" and it says it's lonely here under team members

you're don't even have a team?

Weird you're here as a signed in user.

but I have no prokects

I literally don't know

oh you're signed onto outlook

scary

there you are

somehow it knows I'm at imperial so that solved that

git

a new vs account

do I choose git or the other option

actually yes

dont think so

so I have to register my email first? like create a new account

oh okay

using uni email

you fill out basic information and then the project will be given

Sign on to VSTS

Oh it's not an invite I don't think.

I havent gotten the invite yet

did you send it to our uni emails in the end?

You can add tests based off the board cards. Pretty cool

I have no idea if it's the best to be honest... I'll look into it but we may be forced to use that by the info they send us

yep, but thinking ahead, like once you've done the just getting it running and recording you can join us I guess since server will share tonnes of code

well, it's basically JSON with a standardised set of elements you can use oriented towards creating polygons on a virtual map of earth

The server at this stage is fairly simple other than keeping track of state.

but yeah those will be joined with the server so we can split that

I don't really know what geoJSON is, if you think it's the way to go then go for it... cool, we should plan that a bit then, David and I jotted some stuff down about layout of some classes and created some basics

Check uni email

Also I've invited everyone

server team

Me and Mikhail

which is why I wanna know

I would need to write a script to parse it into Cesium telemetry

also are we agreed on using geoJSON for drone positions an hazards?

same, can't think of anything atm btw who's server team again? The test harness and Server and going to use pretty much entirely the same classes so it'll make more sense to plan that together a bit

how's the c# doing? what's progress?

I need to write the extremely complicated Javascript xDD

I can't think of anything in particular no.

if not thar's fine I'll just work on the VM

no xD I'm just saying if any of you are around we cando stuff

is there anything you guys need me to ask AA? because I'm sending an email by today after the meeting

we need to be available?

I'm in a lecture! will be available around 1

cheers Oliver

sweet, first time I can say bless microsoft

Turns out we might not need Trello at all as VSTS provides scrum boards

I've just set up the project so I'll invite you guys. I think you can use uni accounts for this.

Ok guys so I'm setting up CI and VC with team services now.

That's what I'm guessing if you can't do it directly through cesium

I suspect they just rendered frames individually and strung them together

I need to send an email to the guys at AA I legit have no clue how the fuck they animated their stuff

But you can use a vpn to do it

Last time I tried it had some issues so I didn't bother

You can set up a vpn to do that

I can ssh in but i cant access through browser

That's how the cloud vm works unfortunately

I should mention Mery got a cloud VM working with cesium but for some reason it can only be accessed from within the college

Also, an API for handling REST https://www.asp.net/web-api

Just so you guys know you can use git with Visual Studio Team Services.

current situation: 1. Bad news: Cesium is absolutely fucking humongous and it will definitely take some serious time to learn 2. Good news: All rendering is done client-side so the web-server needs very little resources 3. The simulation takes care of Earth curvature for all polygons 4. DoC VM has an error right now, I've been told by CSG to drop them a ticket and they'll fix it for me as soon as possible 5. I'll be going for non-CSG maintained Ubuntu 16, on which we can get root access, if nobody objects

Be there asap

󾍔󾍔󾍔 don't hate me

https://cesiumjs.org/

GODDAMIT MERY

Im in a lecture 󾰬

I'm gonna go have a coffee before I die but I'll be there straight after

OK

I believe so

So we're meeting at 11 tomorrow, right?

That makes our lives easier

Sweeet

Oh, on the point about CI. Yes VSTS does it. https://www.visualstudio.com/team-services/continuous-integration/

On my way

room 554, fifth floor

everybody remember meeting at 3

tht's completely fine

I'll be helping out at the sponsor's exhibition from 12 so might not be able to make it. I'll let you know.

so anytime after that finishes is fine for me

I'll be in for the robotics lectures / labs

how about we go in at 14:30 and finish it up?

we can run this through the rest tomorrow

ok it looks good to me!

might make me seem like an asshole or something but hey who cares haha

it's completely fair enough Someone needs to correct them though

Looks good. Again, we can look over it tomorrow.

Not surprised by the typos. Just rushed through it.

we should just get a google doc for next time and edit it concurrently

some typos I think and minor rewording but some of it might be opinion

if you agree with them

I made some changes which I highlighted

just reading

Ok guys, here is a draft of the intro and the agile method. Feel free to proof read it.

Exactly

Then write it up properly

Get him to sign off on that

Yeah if we're vague shouldn't take long Could even write a draft in the meeting as notes

We'll probably be discussing it there anyway.

It's not going to take longer than 30 minutes to write iteration 1: do this iteration 2: do that

But we might not have until 7 Because he might go home before then...

We need SOME idea of what we're doing though So I guess we will just have to meet him twice Some point after the meeting

That's why I said to write a basic one.

Since technically that's lying

Yh but he might not sign it without a plan

Which is what I've been saying.

so I guess we'll just have to get him to sign it and make the plan afterwards...

"I certify that I have seen a project plan from my project group that I think is reasonable, given the project brief that they were set. I am happy with where the group plans to get to by the first iteration checkpoint at the end of week 3."

actually looking at the spec supervisor needs to sign off on our plan

stupid that they want us to plan it out even though he said the plan would change but yeah that bit we can't do but the rest we can

My point exactly. Although it will be straight forward once we know what we're doing.

oh right we need to like plan out the full thing right weird I guess 2 week iterations right and then yeah we'll have to do that planning stuff when we know what we're doing but that actually will have to wait yeah

I'll write up the method (scrum) but if you can think of anything to do in terms of iterations before tomorrow afternoon, that would be great.

That and a plan for the iterations (generally what we want to get done at X iteration and why)

that kind of shit

like e.g scrum is a good idea because we'll be working at different times and will need to catch up since we won't know what each other has done

this is just our methods though right Scrum meetings and iterations or xp and that kind of stuff we can plan that without knowing what we're doing for those drones

Like I said we've got a couple of hours which is plenty of time.

But if you want to write a basic plan beforehand, go ahead. All I'm saying is is that it's tricky to write a plan when we have no idea what's going on.

I'm going of what they said in the intro lecture so maybe not perfect info Sounded like they needed to sign the plan That's okay then But I still don't like changing it afterwards We've got time We can do it beforehand

Read the spec

The signature is on a separate sheet

So it'd be hard to change it afterwards since we need a signed copy

Yep exactly my point, need to plan But we need him to sign it right Then we scan it in

We can't write a plan of we don't know what we're doing.

We can get him to approve what we've written, then just write up the rest before the deadline.

But we need to get his approval We'd have to meet him twice and arrange a separate meeting Makes more sense to have it ready then do it in the meeting

yeah we just need to give a method so chatley's happy

After the meeting tomorrow. We have 3 hours.

When are we going to talk about it, or maybe someone can just write it all as bullshit and we can change

Probably. We can write up the plan (doing what for each iteration) after the meeting, though. We don't need to follow it through if we don't want to anyway.

Although length of iteration and how often the meetings are and all that needs to be discussed

True We should discuss it But We're probably just going to do scrum though right

We can't really write the plan without knowing what we're doing. We can get him to sign it before as it's just a formality.

So Need something done before tomorrow Or after the robotics lecture But that's last minute

Yeah we need something for him to approve

yeah I know that far but that's really about it xD

Agile methods. Apparently we need it in the report.

I'll read

I have no idea what any of the 3 are

I might do a little write up tonight on the development method (scrum, kanban, EXTREME) we'll be using for the report tomorrow. I think we agreed on scrum but if anyone else has other ideas let me know.

Cool

P.S. I have emailed Altitude Angels so they are aware of the meeting. They have yet to confirm their attendance, however.

Hi Paul Yes that's right, please come to Room 554, tomorrow 3pm. Then we can decide what each group is going to do, whether something complementary, or some sort of competition . Hopefully someone from Altitude Angels will also join us. Cheers Will

Maybe. We'll see depending on his response.

should I make a group for all 12?

let us see

alright I sent him an email

cc'd though. Email to ask him about it.

it does say "include group 28" though

Though that's for the other group (I assume)

check your email

I'm fine with that too :D As long as the project is good, I'm fine with anything :P

Haha ill make the project amazing but you'll have tl see me im afraid 󾌴

Not me

btw is anyone here doing robotics? we need groups of like 4 or 5

tbh if I never see you but you somehow make our project amazing I'm fine with that too :')

Promise it wont be like this after Thursday haha󾍔󾍔

I know!!!!! Hahaha im sorry about that󾌴󾌴 its been a crazy week

At some point those of us who've never met you will actually get to meet you hahaha

Guys i wont be in on Wednesday for religious reasons (jewish celebration) could you let will know at the meeting? Ill email him anyway but just in case 󾌵 sorry about that and ill catch up with u guys on Thursday if thats okay

Maybe send another email tomorrow? You did send it right before the weekend.

Good start

#radiosilence

tbh ^

I hadn't even heard of the guy until this xD

or have I who is he

yep never met him but he hates me

Mb he doesn't like Bruce

well I'd hope it wouldn't be! never know someone in our group might have a history

Lol why would it be anything personal?

cool least it's nothing personal...

Chatting to Paul and Sam. They haven't heard from him yet. Although that might have changed.

Can you ask the other group if they've got an answer from him? early days but we need to have an approved plan by wednesday right

nope, checked at 10:30

Any update from Will?

I've only just gone around the ground floor entirely. The Assyrian stuff is amazing.

ancient egypt and the rosetta stone were also big hits

oh I've seen it all before but they especially liked the clocks gallery

How far did you get?

I just went to the British Museum with some friends

Sounds interesting

Bots were also cool.

Also went over some docker

It was more showcasing stuff. There's machine learning tools within Azure.

Aren't we using an azure thing Was it useful?

Microsoft had a conference today at Imperial. Went through a lot of their new tech with Azure.

The what?

I was at the Imagine Cloud conference. bretty good

So is anyone doing anything productive? Not having started lectures yet I'm not entirely sure what to do with myself...

Guess we'll find out when we meet with the guy at some point

for monitoring purposes

They might want an interface that's accessible via the browser

I wonder if we can work in anything or whether there's gonna be 1 language which is like a lot easier to use for this and we'll just have to use it

I just never liked it myself. I'm fine using it if I have to, but I'd rather not.

what problems did you run into? I'm curious

xDD

y'all need jesus

Yeah, the same way a knife is useful when your arm is trapped in a rock.

I've always found it useful

but why

Would rather use anything else

I was the javascript guy for my Web apps group Can confirm

what really? it's not that bad

I would only ever use javascript if I literally could not use anything else.

Ew no java script

frankly I'd prefer it xD

unless you want to do javascript

at unity games

and even then you might as well do it in Java

C# is only useful for Steam bots and Windows apps

I did my A level one in C# Do I know any C# now? God no

fun times

I did zero programming until Imperial's Haskell

We did our A level project in Pascal

takes me back to my GCSE days

in visual basic 6

or even better

drone air traffic control in BASIC

that would be bants let's do it drone air traffic control in assembly

I was joking. Fucking hated it.

you're the kind of person that programs in assembly for fun I bet

Actually *****fuck isn't much more complicated than assembly.

lol imagine

better get learning...

well I just meant read up about it lol e.g, if it turns out it's written in BrainFuck

I wouldn't bother xD

the API is password protected

sweet should probably read up on that then

which is an absolutely enormous drone management corporation planning to map everything which flies

the project mentions Altitude Angel

well we're definitely not working from scratch

else we're both just working separately and from scratch

I assume it'll be coordinated though

it might be working on separate features that'd suggest we're not working from like scratch

the project, not Paul xD

this is gonna be one tough nut to crack

so I talked to Paul Balaji and he says he has no clue about the double assignment but that he assumes we'll get to work on separate but related features or something

just so we can easily get the group number...

Amazing!

okay sent

nope I guess we just wanna meet him to get the specifics right

nothing particularly

any specific message to william from any of you?

holy fucking shit x2

We should try to get a meeting before submitting our plan on Wednesday.

We can ask them for help lol

Will's great. This will be actually enjoyable.

holy fucking shit

that's why it said like "of 2 places" or something when we were looking at them

I think projects can get allocated to multiple groups

group 11 AND us???

it says not here

wait what

Good thing we've got 6 people instead of 5

also group leader, arrange a meeting with our supervisor

I can be excited to kill myself? yeah tbh I can go with that

Why not both?

I don't know whether to just kill myself now or be excited

WE GOT THE DRONES ONE

damn well I guess it'll be tomorrow evening then

well, apparently they were too lazy. nothing there yet

I was gonna say today but seems soon considering deadline was yesterday guess we'll find out when we get the email//

I thought it was today. Maybe people are late submitting so tomorrow possibly.

when do we find out which one we got? was it friday some time?

Doesn't look like it.

cool cool

um it doesn't look like it but I legit have no clue

do we need to sign or anything or is it all done?

David and Misha haven't said anything but I doubt they care that much so go for it

should I put it on cate then

yeah hmm I'm fine with the ordering if everyone else is

Also Microsoft support

and quite a bit more impressive on a cv is my guess

but viable

seems much harder

I think most of us did.

bearing in mind it's realistic we could actually get the drones one

although drones and quantified lab were tied, do we like them in that order?

Cool

Full ordering is: 1: drones 2: quantified lab 3: emergency simulator 4: Housing machine learning thing 5: Visble 6: realistic NPC then if Misha, David and Oliver are ok w/ that?

Im okayy too!

yeah im ok with that

well drones and quantified lab are tied too ordering does matter I think, just in case but 4th Housing machine learning thing 5th: Visble 6th - realistic NPC?

I don't think the ordering matters past 3

1. drones 2. quantified lab 3. emergency simulator 4. ?

so ordering is...

alright fine

boo emergency simulator

Quantified lab and NPC win out of mark's ones Drone and Emergency thing are top 2 of the others

Done it!

it's cool once you're done we've got all 6 responses and I'll post them

np

Sorry guys just landed in london

Ill fill it in now

just gimme a ring on 07565849416 if I'm not answering on here

yeah no problem

We don't need you if you've filled out the form. We'll need you just to enter the selections.

alright u guys need me for anything? or can I go home

Cool, Oliver and I have done it as well so Misha and Mery need to do it

I did it too

I did it

or better question, who HAS done it

Who hasn't done it yet?

done

6 o'clock

when's deadline? just doing now

like right now

do it soon ASAP

it's for picking our choices

Do the form

https://goo.gl/forms/0JlQXAEQQZGMQmU53

If you're here come to quiet labs

3 of us in labs atm

Project page was the only thing I signed so I hope not

I signed on the project page, do I have to sign something on cate?

But amongst us 4 earlier the quantified labs and automarker were quite convincing winners

And Mark's ones seem to be disputed

We're gonna meet in person tomorrow to finalise it all after the lecture But atm the 4 not from Mark are pretty tied

Drone thingie mb?

Any other ones stand out?

I like npcs

I think the consensus on the playlist predicting was it's been done before a lot But if you're set on something we Could add it to the list

Misha any thoughts?

Playlist prediction is good too

just for reference

1423 visible 2332 drone atcs 3241 housing market predictions 4114 emergency simulator 10 10 10 10 1231 ml automarker 1222 quantified lab 1343 realistic NPCs 4434 live whiteboard coding 7 7 11 15

Those are mark's as well don't u like them?

I wouldn't chose any of those haha

Attendance capture or automated story illustration from the full list

xD

Sorry

Ooops

live whiteboard coding realistic NPCs ml automarker quantified lab

just these

mark's don't include those!!

Or drones

Emergency simulator from mark's options

also it's a phone app so the programming will have to be learnt from scratch

fitness one seems quite hard to us, how on earth do you figure out if somebody is bench pressing or doing pull-ups with one phone-sized sensor xD

any of mark's seem appealing?

Bob the simple bot - for teaching children to program Fitness Mobile Phone App (Mobile App Development & Machine Learning) Machine Learning] London Housing Price Prediction App Simulating emergency services in cities [Java, City maps]

Have u seen it?

And then anyone from my list

I like drones

Okay

in no particular order

visible drones air traffic control system housnig market predictions emergency simulator

the other 4 fwhich are not mark's which we found interesting:

Plus who doesn't love Mark

Nope But those were the ones we found interesting

And it has to be mark?

Can only have 2 per supervisor so need to choose 2

those are our favourite 4 from mark

Are those the only options?

please sign on projects portal guys :D

Let me read the descriptions

(mikhail and mery)

need to choose max 2

which do you guys like?

these are all from mark

live whiteboard coding realistic NPCs ml automarker quantified lab

OK omw

Main labs

Where are we meeting?

I'm on ray38 in labs, near entrance

wow really, dedication

Yeah, thanks, read all the chat messages

we meet at 14:30 mikhail (just in case you didn't know already)

goody

So we have 6 now assuming everyone stays!

Ill be active on here so do let me know!

Definitely

Will do! I assume you'll be active on here throughout the day though because if you agree with our choices if we make them then we'll need to form the group and stuff on cate

But keep me updated tomorrow

Im not in london until Wednesday! 󾌣

sure see you there :)

Okay so let's do 2:30 then?

2:30 is fine with me.

i mean im not going but for whoever does

arent aci talks at that time?

Does anyone really really wanna stay after 2 or can we say like 2:30?

lmao

Ok I'm fine with anytime after 2 (that's when the food comes in)

Oh wow I had forgotten it was at 4. Maybe we should meet earlier

yay

Cool We can meet in labs by main entrance? & yes David I'm missing the free pizza, I could always get you to steal me some :P Yep 4 That gives us 2 hours to see if we need to find other people or if we agree Which is tight...

so meet at 4 ye?

I'm fine with after ACI

Missing free pizza :o? I'm up for meeting after because that would probably be more productive but it would be good if people could have an idea what things they want to do.

mm I'm not sold, if we can find someone else then I'm more inclined to take them, but if not but yeah so everybody up for meeting tomorrow? Maybe after the ACI talks since I won't go to that

he'll do good

We have till 6pm right so we could meet at or before the talks with free pizza and discuss it

anybody invite him into here?

I think it depends what Tom wants to do, we don't really want people doing something that they are not interested

Sure I'm going to ACI talks anyway (free food)

We could meet tomorrow if everyone is up for it and look over options More will be achieved in person I think As for the other person Idk I want someone not lazy and who doesnt need coercion

Projects I'm interested in are: Simulating emergency services in cities [Java, City maps] [Machine Learning] London Housing Price Prediction App Realistic NPCs Make me a playlist (for fun) Automated Story Illustration

Like I said I'm fine either way. Anyone else?

sooo do I say yes or do I say no

as you can see my luck is generally excellent

he just needs to be coaxed a bit xD

so he's less lazy than most

in wacc we had a guy with asperger's and a girl who did zero, then we had a guy with legit anger issues and another inept chinese person

and the entire wacc

tom and me did the entire pintos

that was our problem

We do need another person but I don't like the sound of lazy.. I've had a lazy group before ( no offense David) and it just meant the rest of the team had to make up the slack

I don't mind either way

everybody ok with him or should I say we have 6 people already?

so I can vouch, but only halfway cause he's a lazy fuck anyway

I've been in 3 groups with him, and in 2 of them we were the only ones working

well, you all have veto right

imo group of 6 implies less work for everyone :D

let's be honest who doesn't?

you guys know Tom Szyszko?

I have a friend asking if he can join

alright guys so we're 5 so far

the security ones are nice

think I'll give this a careful look and respond tonight at some point

We've got until Wednesday afternoon to decide.

im still deciding

If anyone has specific favourites do share them like Mery did I'll have to have a close look at the choices tomorrow

Hmm this will be a tough choice

the medical scan is good too

Yeah people are still working up to the deadline. Btw, we can have up to 6 choices (in ranked order).

There's definitely some interesting stuff there

I think there's more going up as well

wow quite a few

I like the medical scan one as well although it's not ai the using layers of 2d scans to build a 3d image

then admin on the 3rd year group project thing

yeah it's cate then projects portal

wait wheres the list? cate right?

heyo

let me know what u guys think

Bob the simple bot - for teaching children to program Fitness Mobile Phone App (Mobile App Development & Machine Learning) Machine Learning] London Housing Price Prediction App Simulating emergency services in cities [Java, City maps]

here are the ones i liked

I like AI related projects as well :) and im flexible

As do I although I'm also open to other ones there.

so like, for me I wanna do something AI related, preferably, some interesting ones available

if we all agree on what we want to do and whatnot

So this is potentially a group (so far) for the 3rd year project!